T R A N S P O R T G L O B A L P R A C T IACComparable Assessment of Container Port Performance E The Container Port Performance Index 2020 A Comparable Assessment of Container Port Performance The Container Port Performance Index 2020 © 2021 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank, together with external contributions from IHS Markit. The findings, interpretations, and conclusions expressed in this work do not necessar- ily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: The World Bank, 2021. “The Container Port Perfor- mance Index 2020: A Comparable Assessment of Container Port Performance.” World Bank, Wash- ington, DC. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be consid- ered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opin- ions expressed in the adaptation are the sole responsibility of the author or authors of the adapta- tion and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third- party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of compo- nents can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. A Comparable Assessment of Container Port Performance Contents Acknowledgments.................................................................................................................... 5 Abbreviations and Acronyms............................................................................................... 6 Glossary......................................................................................................................................... 7 Foreword........................................................................................................................................ 8 Executive Summary.................................................................................................................10 1.  Introduction........................................................................................................................27 2.  The Measurement of Port Performance.................................................................33 2.1. Introduction..........................................................................................................33 2.2. Measures of Operational or Financial Performance...............................................34 2.3. Measures of Efficiency...........................................................................................37 2.4. Measures Using Data from Exogenous Sources....................................................39 3.  The Approach and Methodology.................................................................................43 3.1. Introduction..........................................................................................................43 3.2. The Anatomy of a Port Call.....................................................................................45 3.3. The Data Underpinning the CPPI...........................................................................49 3.4. The Construction of the CPPI.................................................................................51 Imputing missing values: The administrative approach....................................................60 Imputing missing values: The statistical approach............................................................62 Constructing the index: The administrative approach.......................................................63 Aggregating all ship size groups ......................................................................................64 Constructing the CPPI: The statistical approach.................................................................66 4.  The Container Port Performance Index 2020........................................................69 4.1. Introduction..........................................................................................................69 4.2. The CPPI 2020.......................................................................................................70 4.3. Conclusions and Next Steps..................................................................................84 Appendix A: The CPPI 2020...................................................................................................85 Appendix B: Constructing the CPPI with Factor Analysis.................................... 107 The Container Port Performance Index 2020 4 Box Box. 2.1. Port Development and Competition in Eastern and Southern Africa: Prospects and Challenges...................................................................................................38 Figures Figure B2.1.1. Average Technical Efficiency among ESA Ports, 2008–17..............................38 Figure 3.1. The Anatomy of a Port Call.................................................................................45 Figure 3.2. Average Arrival Times at Hub Ports per Ship Size Range....................................47 Figure 3.3. The Structure of the CPPI...................................................................................51 Figure 3.4. In-Port Time per Port Stay Step or Process.........................................................52 Figure 3.5. Regression Analysis: Crane Intensity versus Call Size..........................................53 Figure 3.6. Relationship between Global Ship Size Index and Call Size Index......................54 Figure 3.7. Gross Crane Productivity or Gross Crane Moves per Hour (GCMPH).................55 Figure 3.8. Average Crane Intensity per Defined Call Size Range.........................................56 Tables Table E.1. The CPPI 2020: Global Ranking of Container Ports..............................................14 Table 2.1. Summary of Performance Indicators...................................................................35 Table 3.1. Disparity in Arrival Times at Hub Ports.................................................................48 Table 3.2. Ports with Low Levels of Matched AIS Data.........................................................50 Table 3.3. Call Size Sensitivity..............................................................................................57 Table 3.4. Ship Size Groups.................................................................................................59 Table 3.5. Call Size and Ship Size Combinations: Percentage of Calls...................................60 Table 3.6. Quantity of Ports Included per Ship Size Group...................................................61 Table 3.7. An Example of Imputing Missing Values.............................................................62 Table 3.8. Port Hours Performance Appraisal......................................................................63 Table 3.9. Assumptions to Determine a Fuel Consumption Index........................................65 Table 3.10. Simple Illustration of Latent Factors..................................................................67 Table 4.1. The CPPI 2020: Global Ranking of Container Ports..............................................71 Table A.1. The CPPI 2020: Administrative Approach.............................................................85 Table A.2. The CPPI 2020: Statistical Approach................................................................... 100 Table A.3. The Ports Omitted from CPPI 2020.................................................................... 106 Table B.1. Sample Port Productivity Data Structure, by Ship Size....................................... 108 Table B.2. Simple Illustration of Latent Factors.................................................................. 110 A Comparable Assessment of Container Port Performance 5 Acknowledgments This technical report was prepared by a joint team from the Transport Global Practice of the Infrastructure Vice-Presidency at the World Bank and the Maritime & Trade division of the consulting firm IHS Markit. The World Bank team was led by Richard Martin Humphreys (Global Lead for Transport Connectivity and Regional Integration and Lead Transport Economist, ITRGK), Ninan Oommen Biju (Senior Port and Maritime Transport Specialist, IEAT1), Hua Tan (Senior Transport Specialist, IEAT2), and Yijie Fei (Consultant, IEAT2), under the guidance of Binyam Reja (Acting Global Practice Director Transport, ITRGK). The IHS Markit team was led by Turloch Mooney (Associate Director, Port Analytics) and Andy Lane (Partner, CTI Consultancy), with the support of Michelle Wong (Research Analyst) and Jim Grausso (Director, Data Transformation), under the guidance of Stuart Strachan (Senior Vice President, Maritime & Trade) and Peter Tirschwell (Vice President, Maritime & Trade). The joint team would also like to extend particular thanks to the following for their comments: Jan Hoffmann (Chief, Trade Logistics Branch, Division on Technology and Logistics, United Nations Conference on Trade and Development, Geneva), Paul E. Kent (Senior Vice President and Managing Director, Ports and Logistics, Monument Economics Group), Jean-Francois Arvis (Senior Transport Economist, ETIRI, World Bank), Juan Samos Tie (Principal Industry Specialist, CNGTR, International Finance Corporation), and Mauricio Posada Baraya (Senior Industry Specialist, CNGTR, International Finance Corporation). The Container Port Performance Index 2020 6 Abbreviations and Acronyms AIS automatic identification system BIMCO Baltic and international Maritime Council CFS container freight station CI crane intensity COVID-19 Coronavirus disease 2019 CPPI 2020 Container Port Performance Index 2020 DBI Doing Business Index FA factor analysis GCMPH gross crane moves per port hour GCI Global Competitiveness Index IAPH International Association of Ports and Harbors ICD inland container depot IMO International Maritime Organization JIT just-in-time JOC Journal of Commerce LDC least developed country LLDC landlocked developing country LPI Logistics Performance Index SIDS small island developing states TEU twenty-foot equivalent unit UNCTAD United Nations Conference on Trade and Development A Comparable Assessment of Container Port Performance 7 Glossary ALL FAST: The point when the ship is fully secured at MOVES PER CRANE: Total moves for a call divided by berth, and all mooring lines are fast. the crane intensity. ARRIVAL: The total elapsed time between the auto- OTHER BERTH HOURS: Activities between all fast and matic identification system (AIS) recorded arrival at the first lift (“start”) plus the time taken to depart from the actual port limit (so excludes waiting time at anchor- berth (all lines released) after the last container lift age) and the vessel all lines fast at the berth. (“finish”). CALL SIZE: number of container moves per call OPERATING HOURS: The time required for container (discharge + Load = Ordered Restows. operations between the first and last container lifts. CARGO OPERATIONS: Total elapsed time between first OTHER PORT HOURS + “ARRIVAL”: The combina- and last container move. tion of idle/waiting time at anchorage plus the time required to steam-in from the port limits and until all CRANE INTENSITY: The quantity of cranes deployed fast alongside the berth. to a ship’s berth call. Calculated as total accumulated gross crane hours divided by operating (first to last PORT CALL: A call to a container port/terminal by a move) hours. container vessel where at least one container was discharged or loaded. FACTOR ANALYSIS (FA): Factor analysis is a statistical method used to describe variability among observed, PORT HOURS: The number of hours a ship spends correlated variables in terms of a potentially lower at/in port, from arrival at the port limits to sailing number of unobserved variables called factors. from the berth. FINISH: Total elapsed time between last container SHIP SIZE: Nominal capacity in twenty-foot equivalent move and all lines released. units, or TEUs. GROSS CRANE HOURS: Aggregated total working time for all cranes deployed to a vessel call without START: The time elapsed from berthing (all lines fast) any deductions. Time includes breakdowns, inclem- to first container move. ent weather, vessel inspired delays, un/lashing, gan- try, boom down/up plus hatch cover and gear-box STEAM-IN TIME: The time required to steam-in from handling and all terminal inspired delays including the port limits and until all fast alongside the berth. meal-breaks. TWENTY-FOOT EQUIVALENT UNIT (TEU): A standard GROSS CRANE PRODUCTIVITY (GCMPH): Call size or metric for container throughput, the number of moves total moves divided by total gross crane hours. loaded and discharged from the vessel, and the physi- cal capacity of a container terminal. A 20-foot container HUB PORT: A “hub port” is defined as a port which is is equal to 1 TEU, and a 40-foot container is equal to called at by deep-sea mainline container ships and 2 TEUs. There is noadjustment for container height. serves as a transhipment point for smaller outlying, or Regardless of container size (10 feet, 15 feet, 20 feet, 30 feeder, ports within its geographical region. Typically, feet, 40 feet, or 45 feet), each is recorded as one move more than 35 percent of its total throughput would be when being loaded or discharged from the vessel. hub and spoke transhipment container activity. VESSEL CAPACITY: Nominal capacity in twenty-foot MOVES: Total container moves. Discharge + ordered equivalent units, or TEUs. restowage moves + load. Excluding hatch covers, gear-boxes, etc. Not adjusted to consider out of gauge WAITING TIME: Total elapsed time from when vessel containers. enters anchorage zone to when vessel departs anchor- age zone (vessel speed must have dropped below 0.5 knots for at least 15 minutes within the zone). The Container Port Performance Index 2020 8 Foreword The COVID-19 pandemic has underlined the critical role that ports, and their associated logistical chains, play in the global economy. It has also high- lighted the need to ensure business continuity and improve the resilience of maritime gateways, as ports act as crucial nodes in the global logistics system, keeping supply chains moving, economies functioning, and people employed. A great variety of public and private stakeholders interact in a port to maintain the flows of vital medical and food supplies, critical agricul- tural products, energy streams, and other goods and services essential to facilitate the economic life of a country. These interactions comprise physi- cal interactions, such as cargo handling operations, vessel-related services, and transfers to and from land-based modes for imports and exports. Maritime transport carries more than 80 percent of global merchandise trade by volume, and any impediment or friction at the port will have tan- gible repercussions for their respective hinterlands and populations. In the short term, this is likely to take the form of shortages of essential goods and higher prices, as we saw early in the pandemic. But over the medium to longer term, an inefficient port will result in slower economic growth, lower employment, and higher costs for importers and exporters. Despite the centrality of the port to global value chains, one of the major challenges to stimulating improvement has been the lack of a reliable, consistent, and comparable basis on which to compare operational performance across different ports. While modern ports collect data for performance purposes, the quality, consistency, and availability of data, the definitions employed, and the capacity and willingness of the organizations to collect and transmit data to a collating body, have all precluded the development of a compara- ble measure (or measures) to assess performance across ports, and time. A Comparable Assessment of Container Port Performance 9 However, the introduction of new technologies, increased digitization, and the willingness on the part of industry interests to work collectively toward systemwide improvements now provides the capacity and the opportu- nity to measure and compare container port performance in a robust and reliable manner for the first time. This technical report, which presents the inaugural edition of the Container Port Performance Index (CPPI 2020), has been produced by the Transport Global Practice of the World Bank, in collaboration with IHS Markit. The CPPI is intended to serve as a reference point for key stakeholders in the global economy, including national gov- ernments, port authorities and operators, development agencies, supra- national organizations, various maritime interests, and other public and private stakeholders in trade, logistics, and supply chain services. As that reference point, the CPPI is intended to identify gaps and opportunities for improvement and hopefully stimulate a dialogue among key stakeholders and move this essential agenda forward. Binyam Reya Stuart Strachan Acting Global Practice Director Senior Vice President, Transport Maritime & Trade The World Bank IHS Markit The Container Port Performance Index 2020 10 Executive Summary 1. Maritime transport is the backbone of globalized trade and the manu- facturing supply chain, with more than four-fifths of global merchan- dise trade (by volume) carried by sea. The maritime sector offers the most economical, energy efficient, and reliable mode of transportation over long distances. A significant and growing portion of that volume, The maritime sec- accounting for approximately 35 percent of total volumes and over 60 tor offers the most percent of commercial value, are carried by containers. The growth of economical, energy containerization has led to vast changes in where and how goods are efficient, and reliable manufactured and processed, an evolving process that has not yet mode of transportation stopped. Container ports, accordingly, are critical nodes in global sup- over long distances. ply chains and central to the growth strategies of many emerging econ- omies. In many cases, the development of high-quality container port infrastructure, operated efficiently, has been a prerequisite to success- ful, export-led growth strategies. It can facilitate investment in produc- tion and distribution systems, supporting the expansion of manufactur- ing and logistics, creating employment, and raising income levels. A Comparable Assessment of Container Port Performance 11 2. Accordingly, how a maritime port performs is a crucial element in the cost of international trade for a country. Unfortunately, ports and terminals, particularly for containers, can often be sources of shipment delays, supply chain disruption, additional costs, and reduced compet- itiveness. Poorly performing ports are characterized by limitations in Unfortunately, ports spatial and operating efficiency, limitations in maritime and landside and terminals, par- access, inadequate oversight, and poor coordination between the pub- ticularly for con- lic agencies involved, resulting in a lack of predictability and reliability. tainers, can often be Poor performance can also have an impact far beyond the hinterland sources of shipment of a port: Container shipping services are operated on fixed sched- delays, supply chain ules with vessel turnaround at each of the ports of call on the route disruption, additional planned within the allocated time for port stay. Poor performance at costs, and reduced one port on the route could disrupt the entire schedule. The result far competitiveness. too often is that instead of facilitating trade, the port increases the cost of imports and exports, reduces the competitiveness of its host country and its hinterland, and inhibits economic growth and poverty reduction. These impacts can be particularly pronounced for landlocked develop- ing countries (LLDCs) and the small island developing states (SIDS). 3. Despite the centrality of the port to global value chains, one of the major challenges to stimulating improvement has been the lack of a reliable, consistent, and comparable basis on which to compare oper- ational performance across different ports. While modern ports collect The introduction of data for performance purposes, the quality, consistency, and availability new technologies, of data, the definitions employed, and the capacity and willingness of increased digitization, the organizations to collect and transmit data to a collating body have and the willingness all precluded the development of a comparable measure (or measures) of industry interests to assess performance across ports, and time. The introduction of new has now provided the technologies, increased digitization, and the willingness of industry opportunity to mea- interests to work collectively toward systemwide improvements has sure and compare now provided the opportunity to measure and compare container port container port per- performance in a robust and reliable manner for the first time. This formance in a robust technical paper, which presents the inaugural edition of the Container and reliable manner Port Performance Index (CPPI 2020), has been produced by the World for the first time. Bank’s Transport Global Practice, in collaboration with IHS Markit. The Container Port Performance Index 2020 12 4. The CPPI is intended to identify gaps and opportunities for improve- ment that will ultimately benefit all stakeholders—from shipping lines to national governments to consumers. The CPPI is intended to serve as a reference point for key stakeholders in the global economy, includ- ing national governments, port authorities and operators, development agencies, supranational organizations, various maritime interests, and other public and private stakeholders in trade, logistics, and supply chain services. The joint team intends that the methodology, scope, and data, will be enhanced in subsequent annual iterations, reflecting The joint team intends refinement, stakeholder feedback, and improvements in data scope that the methodol- and quality. ogy, scope, and data, will be enhanced in 5. The CPPI 2020 has been developed based on total port time in the subsequent annual manner explained in subsequent chapters. This first iteration of CPPI iterations, reflecting utilizes data up to the end of the first six months of 2020 (June 30) and refinement, stake- includes ports that had, within a six-month period in the prior twelve holder feedback, and months, a minimum of 10 valid port calls. The CCPI 2020 was con- improvements in data structed based on two different methodological approaches, or what scope and quality. have been termed the administrative approach: a pragmatic meth- odology reflecting expert knowledge and judgment, and the statis- tical approach, using factor analysis (FA). The rationale of using two approaches was to help ensure the rankings of container port perfor- mance reflected as closely as possible actual port performance, while also being statistically robust. The two approaches are explained later in the report, with further details provided in appendix B. 6. In chapter 4, table E.1 presents the CPPI 2020 using the two meth- odological approaches. The ranking and score in the left-hand col- umns result from the use of the statistical approach and the ranking and score in the right-hand columns result from the administrative approach. The index points used to construct the ranking in the admin- istrative approach reflect the approach outlined in chapter 3 of the report, which is an aggregate of the performance of the port, weighted relative to the average, across call and vessel size. Accordingly, the score can be negative, where a port compares poorly to the average in one call size and vessel size category, particularly if they do not have an offsetting positive score(s) in other cell(s). Further iterations of the CPPI will explore the determinants of the ranking in more detail. A Comparable Assessment of Container Port Performance 13 7. The use of FA also results in a statistic (a total score), that equals the sum of a weighted average of indices for each of the same five vessel sizes. The indices for each vessel size are estimated based on the time expired in the port, and a number of unknown factors, or latent vari- ables (see appendix B for a more detailed explanation of the approach), The resulting total which impact on performance, but cannot be seen. The resulting total scores are standard- scores are standardized, with a “negative” score indicating a better than ized, with a “negative” average performance. Overall, there is a broad consistency between score indicating a the rankings that result from the two approaches, with some excep- better than aver- tions. Just under 18 percent of all ports (61 ports) are ranked within age performance. three places or less from themselves in the two rankings. Approxi- mately 40 percent (137 ports) are ranked with ten places or less of themselves in the respective rankings, while 80 percent (282 ports) fall within 10 percent of their respective rankings in the two indices. 8. The top ranked container ports in the CPPI 2020 are Yokohama port (Japan), in first place, followed by King Abdullah port (Saudi Arabia) in second place. These two ports occupy the same two positions irre- spective of the methodology. The top 50 ranked ports are dominated by ports in East Asia, with ports in the Middle East and North Africa region, such as King Abdullah port, Salalah in Oman (ranked 6th and 9th respectively), Khalifa port in Abu Dhabi (ranked 26th and 22nd The top 50 ranked respectively), and Tanger Med (ranked at 27th and 15th respectively) as ports are dominated the notable exceptions. Algeciras is the highest ranked port in Europe by ports in East Asia. (ranked 10th and 32nd respectively), followed by Aarhus (ranked 44th and 43rd respectively). Colombo is the top-ranked port in South Asia (ranked 17th and 33rd respectively). Lazaro Cardenas the highest ranked port in Latin America (ranked 25th and 23rd respectively), with Halifax the highest ranked port in North America (ranked 39th and 25th respectively). No ports in Sub-Saharan Africa (SSA) rank in the global top 50 container ports; Djibouti port ranks the highest (61st and 93rd respectively). The Container Port Performance Index 2020 14 Table E.1. The CPPI 2020: Global Ranking of Container Ports Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points YOKOHAMA 1 −5.995 YOKOHAMA 1 130 KING ABDULLAH PORT 2 −5.684 KING ABDULLAH PORT 2 114 CHIWAN 3 −5.202 QINGDAO 3 102 GUANGZHOU 4 −5.162 KAOHSIUNG 4 99 KAOHSIUNG 5 −4.669 SHEKOU 5 94 SALALAH 6 −4.531 GUANGZHOU 6 92 HONG KONG, HONG KONG SAR, CHINA 7 −4.276 HONG KONG, HONG KONG SAR, CHINA 7 89 QINGDAO 8 −3.860 ZHOUSHAN 8 88 SHEKOU 9 −3.726 SALALAH 9 87 ALGECIRAS 10 −3.597 YANGSHAN 10 87 BEIRUT 11 −3.378 TANJUNG PELEPAS 11 86 SHIMIZU 12 −3.361 SINGAPORE 12 83 TANJUNG PELEPAS 13 −3.342 NINGBO 13 83 PORT KLANG 14 −3.334 PORT KLANG 14 78 SINGAPORE 15 −3.279 TANGER MEDITERRANEAN 15 76 NAGOYA 16 −3.251 TAIPEI, TAIWAN, CHINA 16 75 COLOMBO 17 −3.209 YANTIAN 17 73 SINES 18 −3.183 CAI MEP 18 68 KOBE 19 −3.127 DALIAN 19 66 ZHOUSHAN 20 −2.963 TIANJIN 20 64 JUBAIL 21 −2.898 FUZHOU 21 61 YEOSU 22 −2.831 KHALIFA PORT 22 60 FUZHOU 23 −2.829 LAZARO CARDENAS 23 60 NINGBO 24 −2.805 SHIMIZU 24 59 LAZARO CARDENAS 25 −2.798 HALIFAX 25 59 KHALIFA PORT 26 −2.795 XIAMEN 26 58 TANGER MEDITERRANEAN 27 −2.769 CHIWAN 27 58 YANGSHAN 28 −2.733 SINES 28 56 A Comparable Assessment of Container Port Performance 15 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points YANTIAN 29 −2.724 MAWAN 29 56 TAIPEI, TAIWAN, CHINA 30 −2.681 AGUADULCE (COLOMBIA) 30 56 DA CHAN BAY TERMINAL ONE 31 −2.588 LIANYUNGANG 31 54 MAWAN 32 −2.557 ALGECIRAS 32 53 DALIAN 33 −2.506 COLOMBO 33 53 INCHEON 34 −2.422 CARTAGENA (COLOMBIA) 34 52 TOKYO 35 −2.418 DA CHAN BAY TERMINAL ONE 35 52 HAMAD PORT 36 −2.411 BUSAN 36 51 LIANYUNGANG 37 −2.375 INCHEON 37 51 PIPAVAV 38 −2.371 HAMAD PORT 38 51 HALIFAX 39 −2.365 PIPAVAV 39 48 CAUCEDO 40 −2.355 YEOSU 40 48 BREMERHAVEN 41 −2.265 AQABA 41 47 CARTAGENA (COLOMBIA) 42 −2.185 JEDDAH 42 46 SALVADOR 43 −2.051 AARHUS 43 43 AARHUS 44 −2.036 MUNDRA 44 43 AGUADULCE (COLOMBIA) 45 −2.035 MAGDALLA 45 42 CAI LAN 46 −1.991 BARCELONA 46 42 HAIPHONG 47 −1.953 RIO GRANDE (BRAZIL) 47 42 MAGDALLA 48 −1.943 NAGOYA 48 41 CAI MEP 49 −1.932 SHANGHAI 49 41 MUNDRA 50 −1.902 KOBE 50 39 GEMLIK 51 −1.892 LAEM CHABANG 51 39 BUSAN 52 −1.887 WILHELMSHAVEN 52 39 JEDDAH 53 −1.862 CHARLESTON 53 38 DILISKELESI 54 −1.842 TOKYO 54 38 LAEM CHABANG 55 −1.807 SALVADOR 55 35 JAWAHARLAL NEHRU PORT 56 −1.786 CALLAO 56 34 AMBARLI 57 −1.783 JUBAIL 57 33 PORT SAID 58 −1.652 YARIMCA 58 33 The Container Port Performance Index 2020 16 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points PECEM 59 −1.647 JEBEL ALI 59 33 AQABA 60 −1.594 HAIPHONG 60 33 DJIBOUTI 61 −1.590 AMBARLI 61 32 XIAMEN 62 −1.541 BEIRUT 62 32 SHANGHAI 63 −1.532 JAWAHARLAL NEHRU PORT 63 31 TANJUNG PRIOK 64 −1.521 KEELUNG 64 31 KEELUNG 65 −1.509 ANTWERP 65 31 TRIPOLI (LEBANON) 66 −1.497 TANJUNG PRIOK 66 31 OSAKA 67 −1.440 WILMINGTON (NORTH CAROLINA, USA) 67 30 YARIMCA 68 −1.393 CAUCEDO 68 30 ITAPOA 69 −1.376 SOHAR 69 29 SANTOS 70 −1.376 PORT SAID 70 29 SOHAR 71 −1.375 BUENAVENTURA 71 29 BUENAVENTURA 72 −1.353 SANTOS 72 28 SEPETIBA 73 −1.338 BOSTON (USA) 73 27 RIO GRANDE (BRAZIL) 74 −1.332 CAI LAN 74 27 KARACHI 75 −1.292 DILISKELESI 75 27 BARCELONA 76 −1.224 TRIPOLI (LEBANON) 76 27 POSORJA 77 −1.192 OSAKA 77 27 OSLO 78 −1.192 BALBOA 78 27 QUY NHON 79 −1.163 GEMLIK 79 26 CAT LAI 80 −1.149 KARACHI 80 26 SUAPE 81 −1.129 YOKKAICHI 81 26 LONDON 82 −1.117 COLON 82 25 PHILADELPHIA 83 −1.080 MERSIN 83 25 DANANG 84 −1.053 SEPETIBA 84 24 PORT OF VIRGINIA 85 −1.044 BREMERHAVEN 85 24 ANTWERP 86 −1.011 PECEM 86 23 ZEEBRUGGE 87 −0.988 PHILADELPHIA 87 23 SANTA CRUZ DE TENERIFE 88 −0.970 ITAPOA 88 23 A Comparable Assessment of Container Port Performance 17 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points NEW YORK & NEW JERSEY 89 −0.969 PIRAEUS 89 23 GDYNIA 90 −0.942 ROTTERDAM 90 22 SHANTOU 91 −0.935 OSLO 91 20 NAHA 92 −0.883 DAMMAM 92 20 PIRAEUS 93 −0.859 DJIBOUTI 93 19 PUERTO LIMON 94 −0.858 PAITA 94 19 CHARLESTON 95 −0.820 SANTA CRUZ DE TENERIFE 95 19 PORT AKDENIZ 96 −0.820 SAVANNAH 96 18 TAICHUNG 97 −0.814 VERACRUZ 97 18 ULSAN 98 −0.811 ALTAMIRA 98 17 HAKATA 99 −0.758 JACKSONVILLE 99 17 COLON 100 −0.752 TAURANGA 100 17 MOBILE 101 −0.745 MARSAXLOKK 101 16 DAMMAM 102 −0.737 VALPARAISO 102 16 PUERTO BARRIOS 103 −0.731 GDYNIA 103 16 NOUMEA 104 −0.721 QUY NHON 104 16 PAITA 105 −0.705 ITAJAI 105 16 MARSAXLOKK 106 −0.698 PUERTO LIMON 106 16 YOKKAICHI 107 −0.692 PARANAGUA 107 16 MALAGA 108 −0.690 JOHOR 108 15 OMAEZAKI 109 −0.680 CAT LAI 109 15 JOHOR 110 −0.669 PORT OF VIRGINIA 110 15 MOJI 111 −0.663 NAHA 111 15 BATANGAS 112 −0.663 FORT−DE−FRANCE 112 15 BOSTON (USA) 113 −0.631 POINTE−A−PITRE 113 14 BALBOA 114 −0.625 MIAMI 114 14 SIAM SEAPORT 115 −0.613 DANANG 115 13 ROTTERDAM 116 −0.611 HAKATA 116 12 BURGAS 117 −0.611 SHANTOU 117 12 DUNKIRK 118 −0.611 AUCKLAND 118 12 The Container Port Performance Index 2020 18 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points WILHELMSHAVEN 119 −0.598 BALTIMORE (USA) 119 12 SANTA MARTA 120 −0.589 DAKAR 120 12 TARRAGONA 121 −0.588 OMAEZAKI 121 11 CAGAYAN DE ORO 122 −0.582 PUERTO BARRIOS 122 11 AUCKLAND 123 −0.562 MOJI 123 11 NORRKOPING 124 −0.553 NOUMEA 124 11 FORT−DE−FRANCE 125 −0.543 ENSENADA 125 11 GUSTAVIA 126 −0.528 POSORJA 126 10 QINZHOU 127 −0.523 NEW YORK & NEW JERSEY 127 10 FREDERICIA 128 −0.493 CAGAYAN DE ORO 128 10 COPENHAGEN 129 −0.471 MALAGA 129 10 TUTICORIN 130 −0.469 SUAPE 130 10 PUERTO CORTES 131 −0.468 GDANSK 131 10 PORT BRONKA 132 −0.454 MOBILE 132 10 TANJUNG EMAS 133 −0.433 BURGAS 133 9 CHIBA 134 −0.418 TARRAGONA 134 9 SHIBUSHI 135 −0.382 NORRKOPING 135 9 ENSENADA 136 −0.378 SIAM SEAPORT 136 9 SAIGON 137 −0.375 SANTA MARTA 137 9 BALTIMORE (USA) 138 −0.368 FREDERICIA 138 9 PENANG 139 −0.367 TUTICORIN 139 8 KHALIFA BIN SALMAN 140 −0.356 CEBU 140 8 PUERTO BOLIVAR (ECUADOR) 141 −0.347 WELLINGTON 141 8 GIOIA TAURO 142 −0.344 COPENHAGEN 142 8 DAVAO 143 −0.341 TANJUNG EMAS 143 8 CASTELLON 144 −0.339 BATANGAS 144 8 ALTAMIRA 145 −0.324 MANZANILLO (MEXICO) 145 8 RIO HAINA 146 −0.322 TAICHUNG 146 8 POINTE−A−PITRE 147 −0.319 PORT EVERGLADES 147 8 A Comparable Assessment of Container Port Performance 19 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points SAN JUAN 148 −0.303 QINZHOU 148 8 SHUAIBA 149 −0.301 PUERTO CORTES 149 7 NAPLES 150 −0.280 CHIBA 150 7 CEBU 151 −0.275 SAIGON 151 7 FREEPORT (BAHAMAS) 152 −0.271 PORT BRONKA 152 7 TANJUNG PERAK 153 −0.269 HAIFA 153 6 PUERTO QUETZAL 154 −0.269 PORT AKDENIZ 154 6 WILMINGTON (NORTH CAROLINA, USA) 155 −0.237 PUERTO BOLIVAR (ECUADOR) 155 6 VIGO 156 −0.233 DAVAO 156 6 PAPEETE 157 −0.227 CASTELLON 157 6 VERACRUZ 158 −0.219 CORONEL 158 6 SAN ANTONIO 159 −0.218 VIGO 159 5 SHUWAIKH 160 −0.216 RIO HAINA 160 5 JACKSONVILLE 161 −0.215 GUSTAVIA 161 5 CALDERA (COSTA RICA) 162 −0.206 KHALIFA BIN SALMAN 162 5 BELL BAY 163 −0.205 BELL BAY 163 5 HELSINGBORG 164 −0.196 SAN JUAN 164 5 POINT LISAS PORTS 165 −0.182 CALDERA (COSTA RICA) 165 4 BORUSAN 166 −0.163 SHIBUSHI 166 4 KALININGRAD 167 −0.159 PAPEETE 167 4 CARTAGENA (SPAIN) 168 −0.153 SALERNO 168 3 BARRANQUILLA 169 −0.137 BARRANQUILLA 169 3 CRISTOBAL 170 −0.135 YUZHNY 170 3 LATAKIA 171 −0.135 HELSINGBORG 171 3 PALERMO 172 −0.125 TANJUNG PERAK 172 3 NASSAU 173 −0.109 SAN ANTONIO 173 3 SALERNO 174 −0.080 PUERTO QUETZAL 174 2 CIVITAVECCHIA 175 −0.072 ULSAN 175 2 WELLINGTON 176 −0.049 RAVENNA 176 2 The Container Port Performance Index 2020 20 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points DAKAR 177 −0.031 POINT LISAS PORTS 177 2 HERAKLION 178 −0.030 LIMASSOL 178 1 RAVENNA 179 −0.025 CARTAGENA (SPAIN) 179 1 CATANIA 180 −0.021 LONDON 180 1 LARVIK 181 −0.015 CIVITAVECCHIA 181 1 PORT FREEPORT 182 −0.008 PALERMO 182 1 VILA DO CONDE 183 0.000 ANCONA 183 0 VITORIA 183 0.000 LARVIK 184 0 OITA 185 0.004 HERAKLION 185 0 KRISTIANSAND 186 0.008 LYTTELTON 186 0 GIJON 187 0.013 LATAKIA 187 0 LYTTELTON 188 0.034 OITA 188 0 ODESSA 189 0.046 NASSAU 189 0 DURRES 190 0.057 CATANIA 190 0 MARIEL 191 0.057 HAMBURG 191 0 LIMASSOL 192 0.063 PHILIPSBURG 192 0 ANCONA 193 0.067 PORT FREEPORT 193 0 PHILIPSBURG 194 0.090 KRISTIANSAND 194 −1 MUUGA—PORT OF TALLINN 195 0.092 BORUSAN 195 −1 VARNA 196 0.094 GIJON 196 −1 BELAWAN 197 0.104 MARIEL 197 −1 YUZHNY 198 0.105 MUUGA—PORT OF TALLINN 198 −1 RIO DE JANEIRO 199 0.126 SHUWAIKH 199 −1 MIAMI 200 0.149 TAMPA 200 −1 BARI 201 0.167 SANTO TOMAS DE CASTILLA 201 −2 VLADIVOSTOK 202 0.182 BELAWAN 202 −2 SANTO TOMAS DE CASTILLA 203 0.185 BARI 203 −2 VOSTOCHNY 204 0.192 TOMAKOMAI 204 −2 TOMAKOMAI 205 0.194 MATADI 205 −2 A Comparable Assessment of Container Port Performance 21 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points LEIXOES 206 0.201 NELSON 206 −2 ARICA 207 0.211 NAPLES 207 −2 PORT EVERGLADES 208 0.213 RAUMA 208 −3 MATADI 209 0.238 POTI 209 −3 NELSON 210 0.240 LEIXOES 210 −3 SAINT JOHN 211 0.241 PORT MORESBY 211 −4 TRIESTE 212 0.241 GIOIA TAURO 212 −4 GAVLE 213 0.242 VARNA 213 −4 MERSIN 214 0.275 GAVLE 214 −4 POTI 215 0.278 NEW MANGALORE 215 −4 TEESPORT 216 0.303 FREETOWN 216 −5 PORT MORESBY 217 0.307 ABIDJAN 217 −5 TIANJIN 218 0.310 VITORIA 218 −5 TAURANGA 219 0.333 TOAMASINA 219 −5 FREETOWN 220 0.350 KALININGRAD 220 −5 SAN VICENTE 221 0.361 ALEXANDRIA (EGYPT) 221 −5 VALPARAISO 222 0.365 GRANGEMOUTH 222 −6 BERBERA 223 0.368 RIGA 223 −6 BILBAO 224 0.384 FREEPORT (BAHAMAS) 224 −6 HAIFA 225 0.384 ZEEBRUGGE 225 −6 TAMPA 226 0.411 SAN VICENTE 226 −7 TOAMASINA 227 0.433 PENANG 227 −7 ABIDJAN 228 0.439 SHUAIBA 228 −7 RAUMA 229 0.439 PANJANG 229 −7 CHORNOMORSK 230 0.449 KOPER 230 −7 CALLAO 231 0.450 DURRES 231 −7 NEW MANGALORE 232 0.451 APRA HARBOR 232 −7 UMM QASR 233 0.452 COTONOU 233 −8 ALEXANDRIA (EGYPT) 234 0.482 BRISBANE 234 −8 The Container Port Performance Index 2020 22 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points RIGA 235 0.482 TAKORADI 235 −8 SEATTLE 236 0.487 PUERTO PROGRESO 236 −8 VANCOUVER (CANADA) 237 0.499 NEW ORLEANS 237 −9 CONAKRY 238 0.505 TEESPORT 238 −9 EL DEKHEILA 239 0.509 BLUFF 239 −9 NAPIER 240 0.513 VLADIVOSTOK 240 −9 LA SPEZIA 241 0.548 VILA DO CONDE 241 −9 KOTKA 242 0.558 HOUSTON 242 −9 GRANGEMOUTH 243 0.560 BILBAO 243 −10 COTONOU 244 0.562 KOTKA 244 −10 OWENDO 245 0.563 BERBERA 245 −10 BRISBANE 246 0.569 NANTES−ST NAZAIRE 246 −10 PANJANG 247 0.573 ARICA 247 −11 APRA HARBOR 248 0.610 PORT OF SPAIN 248 −11 ACAJUTLA 249 0.640 ACAJUTLA 249 −11 DAMIETTA 250 0.650 CONAKRY 250 −11 LIVORNO 251 0.658 SAINT JOHN 251 −12 PARANAGUA 252 0.659 DUBLIN 252 −12 CONSTANTZA 253 0.660 STOCKHOLM 253 −13 TAKORADI 254 0.683 MALABO 254 −14 NANTES−ST NAZAIRE 255 0.693 VOSTOCHNY 255 −14 LIRQUEN 256 0.708 EL DEKHEILA 256 −14 BLUFF 257 0.731 CRISTOBAL 257 −14 LE HAVRE 258 0.746 ST PETERSBURG 258 −14 PUERTO PROGRESO 259 0.750 BANGKOK 259 −15 PORT OF SPAIN 260 0.775 MOGADISCIO 260 −15 TEMA 261 0.782 NAPIER 261 −15 DUBLIN 262 0.839 LIRQUEN 262 −15 MALABO 263 0.859 SEATTLE 263 −16 A Comparable Assessment of Container Port Performance 23 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points HUENEME 264 0.862 TRIESTE 264 −16 NEW ORLEANS 265 0.865 ODESSA 265 −17 HOUSTON 266 0.878 KLAIPEDA 266 −17 STOCKHOLM 267 0.915 AGADIR 267 −17 BEIRA 268 0.919 CASABLANCA 268 −17 MANAUS 269 0.967 IZMIR 269 −18 ONNE 270 0.994 MANAUS 270 −18 BANGKOK 271 1.024 OWENDO 271 −18 ISKENDERUN 272 1.024 ISKENDERUN 272 −18 MONTEVIDEO 273 1.033 HUENEME 273 −19 VENICE 274 1.042 CHORNOMORSK 274 −19 MANZANILLO (MEXICO) 275 1.070 BATUMI 275 −19 AGADIR 276 1.122 MANILA 276 −19 IZMIR 277 1.136 BRISTOL 277 −19 GENERAL SANTOS 278 1.148 LAE 278 −19 SAVANNAH 279 1.158 RIJEKA 279 −20 HAMBURG 280 1.176 SAMSUN 280 −20 MOGADISCIO 281 1.194 BEIRA 281 −21 CORONEL 282 1.203 GENERAL SANTOS 282 −21 VALENCIA 283 1.211 MONTREAL 283 −22 THESSALONIKI 284 1.229 THESSALONIKI 284 −22 MONTREAL 285 1.231 CONSTANTZA 285 −22 KINGSTON (JAMAICA) 286 1.283 KINGSTON (JAMAICA) 286 −23 LAE 287 1.314 KRIBI DEEP SEA PORT 287 −24 LOME 288 1.332 OTAGO HARBOUR 288 −24 MEJILLONES 289 1.381 LA SPEZIA 289 −24 SOUTHAMPTON 290 1.404 LIVORNO 290 −24 CASABLANCA 291 1.442 ONNE 291 −25 PORT VICTORIA 292 1.457 NOUAKCHOTT 292 −25 The Container Port Performance Index 2020 24 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points SOKHNA 293 1.459 SAN PEDRO (COTE D'IVOIRE) 293 −27 BRISTOL 294 1.462 UMM QASR 294 −27 NOUAKCHOTT 295 1.475 VENICE 295 −27 MUHAMMAD BIN QASIM 296 1.499 LA GUAIRA 296 −28 SAMSUN 297 1.502 DAMIETTA 297 −28 DOUALA 298 1.510 CHATTOGRAM 298 −28 MAPUTO 299 1.533 PORT VICTORIA 299 −28 NOVOROSSIYSK 300 1.626 LE HAVRE 300 −30 KLAIPEDA 301 1.635 NOVOROSSIYSK 301 −30 MELBOURNE 302 1.676 DOUALA 302 −31 KRIBI DEEP SEA PORT 303 1.701 VANCOUVER (CANADA) 303 −31 ST PETERSBURG 304 1.719 TIMARU 304 −32 LA GUAIRA 305 1.737 DUNKIRK 305 −33 CHATTOGRAM 306 1.809 MAPUTO 306 −33 WALVIS BAY 307 1.819 PUERTO CABELLO 307 −34 ITAJAI 308 1.828 VALENCIA 308 −34 PUERTO CABELLO 309 1.829 RIO DE JANEIRO 309 −35 SAVONA−VADO 310 1.897 BUENOS AIRES 310 −37 GDANSK 311 1.944 MEJILLONES 311 −38 RIJEKA 312 1.974 TEMA 312 −40 FELIXSTOWE 313 2.006 MELBOURNE 313 −40 OTAGO HARBOUR 314 2.023 SAVONA−VADO 314 −40 ALGIERS 315 2.076 ASHDOD 315 −42 BATUMI 316 2.175 MUHAMMAD BIN QASIM 316 −42 TIMARU 317 2.225 SOUTHAMPTON 317 −45 KOPER 318 2.237 ALGIERS 318 −49 SAN PEDRO (COTE D'IVOIRE) 319 2.267 FREMANTLE 319 −49 BUENOS AIRES 320 2.391 IQUIQUE 320 −54 GENOA 321 2.420 SOKHNA 321 −55 MANILA 322 2.445 FELIXSTOWE 322 −55 A Comparable Assessment of Container Port Performance 25 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points JEBEL ALI 323 2.482 PRINCE RUPERT 323 −56 DAR ES SALAAM 324 2.561 DAR ES SALAAM 324 −58 DUTCH HARBOR 325 2.591 DUTCH HARBOR 325 −60 FREMANTLE 326 2.716 NEMRUT BAY 326 −61 ASHDOD 327 2.797 PORT BOTANY 327 −63 LOS ANGELES 328 2.899 MONTEVIDEO 328 −65 NEMRUT BAY 329 2.970 TACOMA 329 −66 PRINCE RUPERT 330 2.979 BEJAIA 330 −72 MOMBASA 331 3.140 GENOA 331 −74 OAKLAND 332 3.163 LOME 332 −77 LONG BEACH 333 3.175 ADELAIDE 333 −78 PORT REUNION 334 3.302 OAKLAND 334 −79 TACOMA 335 3.628 MOMBASA 335 −79 GUAYAQUIL 336 3.647 WALVIS BAY 336 −80 PORT BOTANY 337 3.907 LOS ANGELES 337 −82 BEJAIA 338 4.054 GUAYAQUIL 338 −84 ADELAIDE 339 4.546 GOTHENBURG 339 −87 LAGOS (NIGERIA) 340 4.646 PORT REUNION 340 −89 GOTHENBURG 341 4.653 LONG BEACH 341 −96 IQUIQUE 342 4.766 LAGOS (NIGERIA) 342 −114 TIN CAN ISLAND 343 4.789 LUANDA 343 −115 PORT LOUIS 344 5.501 TIN CAN ISLAND 344 −118 MARSEILLE 345 5.696 POINTE−NOIRE 345 −128 POINTE−NOIRE 346 5.832 PORT LOUIS 346 −175 CAPE TOWN 347 6.528 CAPE TOWN 347 −177 PORT ELIZABETH 348 7.659 PORT ELIZABETH 348 −183 DURBAN 349 8.082 NGQURA 349 −190 LUANDA 350 8.383 MARSEILLE 350 −238 NGQURA 351 8.401 DURBAN 351 −255 The Container Port Performance Index 2020 26 9. While the rankings demonstrate a broad consistency resulting from the two approaches, some exceptions exist. Three of the most notable discrepancies are Jebel Ali (323rd and 59th in the statistical and admin- istrative approach respectively), Itajai (308th and 105th respectively), and Tianjin, one of the highest ranked ports affected (218th and 20th respectively). The approach taken in the CPPI 2020 has been not to try and explain every discrepancy, but rather to make the methodology and assumptions explicit and let the data speak. In future editions, further refinement of both approaches will hopefully reduce discrep- ancies, and further enhance their respective complementarity, and the The overall objective rigor of the CPPI. remains the iden- tification of poten- 10. Looking to the future, the intention is that the CPPI will evolve and be tial improvement to enhanced in subsequent editions, reflecting refinement, stakeholder ultimately benefit feedback, and improvement in data scope and quality. The WBG/ all public and pri- IHS Markit team will continue to refine the methodologies, the scope, vate stakeholders. where possible increasing the number of ports, and the data. The next iteration, CPPI 2021, will be comparable, facilitating the introduction of trends in container port performance, both overall and potentially by disaggregation by ship or call size. The CPPI 2021 will also seek to investigate and explain divergences between the two approaches, while also gaining a further understanding of key determinants or influences on container port performance. The overall objective remains the iden- tification of potential improvement to ultimately benefit all public and private stakeholders including ports, shipping lines, governments, line agencies, businesses, and consumers. A Comparable Assessment of Container Port Performance 27 1. Introduction 11. Maritime transport is the backbone of globalized trade and the man- ufacturing supply chain, with more than four-fifths of global mer- chandise trade (by volume) carried by sea. The maritime sector offers the most economical and reliable mode of transportation over long distances, with volumes carried increasing at an annual average of 3 percent; over the period from 1970 to 2018. In 2019, maritime trade volumes expanded by only 0.5 percent, reaching a total of just over 11 billion tons for the year (UNCTAD 2020). Container trade, which accounts for approximately 35 percent of total volume and for more than 60 percent by value, grew by 2 percent, down from 5.1 percent in 2018. Overall, both maritime and container trade reflect a slowdown in world gross domestic product (GDP) to 2.5 percent, 0.6 percent lower than the 3.1 percent GDP in 2018. Lingering trade tensions and high policy uncertainty undermined growth in global economic output and merchandise trade, and by extension, maritime trade. The Container Port Performance Index 2020 28 12. Since the dawn of maritime trade, maritime ports have been central to economic and social development. This statement is as true today as it has been for thousands of years. The growth of containerization, since Malcom McLean’s innovation in 1958 (Levinson 2006), has led to Container ports have vast changes in where and how goods are manufactured, a dynamic become critical nodes process that has continued shaping the industry. Container ports, as a in global supply chains result, have become critical nodes in global supply chains and are cen- and are central to the tral to the growth stories and strategies of many emerging economies. growth stories and In many cases, the development of high-quality port infrastructure, strategies of many operated efficiently, has served as a prerequisite to successful, often emerging economies. export-led, growth strategies. Done well, port infrastructure provides the necessary confidence to facilitate investment in production and distribution systems, supporting the expansion of manufacturing and logistics, while creating employment and raising income levels. 13. More specifically, how a maritime port performs is a crucial element in determining a country’s trade costs. Poorly performing ports con- strain trade growth, an impact particularly pronounced for landlocked developing countries (LLDCs) and the small island developing states (SIDS). The port, together with the access infrastructure (whether inland waterway, rail, or road) to the hinterland constitutes a crucial link to the Poorly perform- global marketplace and needs to operate efficiently. Efficient perfor- ing ports constrain trade growth. mance encompasses a myriad of factors, including the efficiency of the port itself, the availability of sufficient draught, quay, and dock facilities, the quality of the connections to road and rail services, the competi- tiveness of those services, and the efficacy of the procedures employed by the public agencies involved in container clearance. Inefficiencies or nontariff barriers in any of these actors will result in higher costs, reduced competitiveness, and lower trade (Kathuria 2018). 14. More specifically, the efficiency of port infrastructure has also been identified as a key contributor to overall port competitiveness and international trade costs. Sanchez et al. (2003) identified a link between port efficiency and the cost of international trade. Clark et al. (2004) found a reduction in a country’s trade inefficiencies, spe- cifically transport costs, from the 25th to 75th percentile, resulted in an increase in bilateral trade of around 25 percent. Wilmsmeier et al. (2006) confirmed the impact of port performance on international trade costs, finding that doubling port efficiency in a pair of ports had the same impact on trade costs as would halving the physical distance A Comparable Assessment of Container Port Performance 29 between the ports. Hoffmann et al. (2020) analyzed the short- and long- run impacts of the liner shipping bilateral connectivity on South Africa’s trade flows, and showed that GDP, the number of common direct con- nections, and the level of competition have a positive and significant effect on trade flows. 15. Unfortunately, ports and terminals, particularly for containers, can Far too often instead often be the main sources of shipment delays, supply chain disruption, of facilitating trade, additional costs, and reduced competitiveness. Poorly performing the port increases ports are characterized by limitations in spatial and operating effi- the cost of imports ciency, limitations in maritime and landside access, inadequate over- and exports, reduces sight, and poor coordination between the public agencies involved, competitiveness, resulting in a lack of predictability and reliability. The result far too and inhibits eco- often is that instead of facilitating trade, the port increases the cost of nomic growth and imports and exports, reduces competitiveness, and inhibits economic poverty reduction. growth and poverty reduction. The effect on national and regional economies can be severe (see World Bank 2013) and has driven numer- ous efforts to improve performance to strengthen competitiveness. 16. Port performance is also a key consideration for container shipping lines that operate liner services on fixed schedules, based on fixed port turnaround times. Delays at any of the scheduled ports of call along the route served by the vessel would have to be resolved before the vessel arrives at the next port of call, in order to avoid an adverse impact on the efficiency of service operations. As such, port efficiency and port turnaround time on a vessel’s scheduled voyage at all ports of Port efficiency and call are important subjects for operators. In addition, monitoring port port turnaround time performance has become an increasingly important undertaking in the at all ports of call competitive landscape. are important sub- jects for operators. 17. One of the major challenges to realize improvement has been the lack of reliable measures to compare operational performance across different ports. The old management idiom: “You cannot manage what you cannot measure,” reflects the historical challenge of both man- aging and overseeing the maritime transport sector. While modern ports collect data for performance purposes, it is difficult to compare the outcomes with competitors or with ports in similar circumstances. Managers might know performance is improving year on year, but they might not know whether performance is up to the standards of leading ports with similar profiles. Not surprisingly, the sector has amassed a The Container Port Performance Index 2020 30 long history of attempts to identify and implement a comparative set of indicators to measure port or terminal performance; a brief review of the related literature is provided in chapter 2. However, one of the general challenges of nearly all these approaches has been the quality, consistency, and availability of data, the standardization of definitions employed, and the capacity and willingness of the relevant organiza- tions to collect and transmit the data to a collating body. 18. At a slightly higher level, several aggregate indicators provide an indication of the comparative quality and performance of maritime New technologies, gateways. The World Bank Logistics Performance Index, or LPI (Arvis et increased digitali- al. 2018), and Doing Business Index, or DBI (World Bank 2020), on trad- zation, and industry ing across borders, and the World Economic Forum’s Global Competi- stakeholder willing- tiveness Index (GCI) 4.0 all report on the perceived efficiency of seaport ness to work collec- tively toward system- services and border clearance processes. In addition, each indicates wide improvements the extent to which inefficiencies at a nation’s sea borders can impact now, for the first time, international trade competitiveness. But the aggregate nature of the provides the capac- indicators, and the fact they are perception based, means they offer at ity and opportunity to best an indication of comparative performance with little to guide spa- measure and compare tial or operating performance improvements at the level of the individ- container port per- ual port. The Liner Shipping Connectivity Index (LSCI) published by the formance in a robust United Nations Conference on Trade and Development (UNCTAD) pro- and reliable manner. vides an indicator or a port’s position within the liner shipping network. While a port’s position is in part determined by the port’s performance, the LSCI does not directly measure it. Similar in scope to the CPPI, the LSCI is also limited to container ports. 19. Digitalization offers an opportunity to measure and compare con- tainer port performance in a robust and reliable manner. New tech- nologies, increased digitalization, and industry stakeholder willingness to work collectively toward systemwide improvements now, for the first time, provides the capacity and opportunity to measure and compare container port performance in a robust and reliable manner. The data used to compile the CPPI 2020 are gathered from the IHS Markit Port Performance Program.1 The Port Performance Program was started in 2009 with the goal of driving efficiency improvements in container port operations and supporting programs to optimize port calls. The pro- gram includes 10 of the world’s largest liner shipping companies, which collectively operate 76 percent of global fleet capacity. The nature, source, and scope of the data are discussed in chapter 3. A Comparable Assessment of Container Port Performance 31 20. The rationale behind the CPPI was to use available empirical data to create an objective measure to compare container port performance across ports, and eventually over time. Container port performance is most relevant from the perspective of customer experience and the speed and efficiency with which customer assets are handled. In this inaugural edition of the CPPI, the focus is purely on quayside perfor- Container port perfor- mance to be reflective of the experience of a ship operator, the port’s mance is most rel- main customer. The operational efficiency with which ports receive and evant from the per- turn around container ships is of critical importance to the carrier’s spective of customer decision to call a port in view of other options. The two methodologies experience and the employed, and the justification for their use, are presented in summary speed and efficiency in chapter 3, with further detail provided in appendix B. The results in with which customer terms of the CPPI 2020 are presented in chapter 4, with detailed tables assets are handled. provided in appendix A. 21. The CPPI has been developed to contribute to the identification of opportunities for improvement that will ultimately benefit all public and private stakeholders. The CPPI is intended to serve as a reference point for key stakeholders in the global economy including national governments, port authorities and operators, development agencies, supra-national organizations, various maritime interests, and other public and private stakeholders engaged in trade, logistics and supply chain services. The intention of the joint team is that the methodology, scope, and data, will be enhanced in subsequent annual iterations, reflecting refinement, stakeholder feedback, and data scope and qual- ity improvements. 1. For more information on the Port Performance Program, see: NOTE https://ihsmarkit.com/products/port-performance.html. The Container Port Performance Index 2020 32 REFERENCES Arvis, Jean-François, Lauri Ojala, Christina Wiederer, Ben Shepherd, Anasuya Raj, Karlygash Dairabayeva, and Tuomas Kiiski. 2018. Connecting to Compete 2018: Trade Logistics in the Global Economy. World Bank, Wash- ington D.C. https://openknowledge.worldbank.org/10986/29971. Clark, Ximena, David Dollar, and Alexander Micco. 2004. “Port Efficiency, Maritime Transport Costs, and Bilateral Trade.” Journal of Development Economics 75 (2): 417–50. https://doi.org/10.1016/j.jdeveco.2004.06.005. Hoffmann, Jan, Naima Saeed, and Sigbjørn Sødal. 2020. “Liner Shipping Bilateral Connectivity and Its Impact on South Africa’s Bilateral Trade Flows.” Maritime Economics & Logistics 22 (3): 473–99. https://doi.org/10.1057/s41278-019-00124-8. Kathuria, Sanjay. 2018. A Glass Half Full: The Promise of Regional Trade in South Asia. Washington D.C.: World Bank. https://openknowledge.worldbank.org/handle/10986/30246 License: CC BY 3.0 IGO. Levinson, Marc. 2006. The Box: How the Shipping Container Made the World Smaller and the World Economy Bigger. Princeton University Press. https://doi.org/10.2307/j.ctvcszztg. Sánchez, Ricardo, Jan Hoffmann, Alejandro Micco, Georgina Pizzolitto, Martín Sgut, and Gordon Wilmsmeier. 2003. “Port Efficiency and International Trade: Port Efficiency as a Determinant of Maritime Transport Costs.” Maritime Economics & Logistics 5 (2): 199–218. https://doi.org/10.1057/palgrave.mel.9100073. UNCTAD (United Nations Conference on Trade and Development). 2020. Review of Maritime Transport 2020. Geneva: UNCTAD. https://unctad.org/webflyer/review-maritime-transport-2020. Wilmsmeier, Gordon, Jan Hoffmann, and Ricardo Sanchez. 2006. “The Impact of Port Characteristics on Interna- tional Maritime Trade Costs.” Research in Transportation Economics 16 (1): 117–40. https://doi.org/10.1016/S0739-8859(06)16006-0. World Bank. 2013. Tanzania Economic Update #3: Opening the Gates: How the Port of Dar es Salaam Can Transform Tanzania. https://www.worldbank.org/en/country/tanzania/publication/ opening-the-gates-how-the-port-of-dar-es-salaam-can-transform-tanzania-backup. World Bank. 2020. Doing Business 2020: Comparing Business Regulation in 190 Economies. Doing Business Series. Washington, DC: World Bank. http://documents.worldbank.org/curated/en/688761571934946384/ Doing-Business-2020-Comparing-Business-Regulation-in-190-Economies. A Comparable Assessment of Container Port Performance 33 2.  The Measurement of Port Performance 2.1. Introduction 22. Chapter 1 underlined the importance of a reliable comparative mea- sure to assess performance across different ports. Perhaps not sur- prisingly, the attempts to identify a comparative set of indicators to measure port or terminal performance stretch back several decades. This chapter offers a brief review of that literature, and it is certainly not intended to be exhaustive. Rather, the intention is to illustrate the broad approaches identified in the literature and comment on the merits and demerits of each. The review has been structured around a simple typology: firstly, measures of operational and financial perfor- mance; secondly, measures of economic efficiency; and thirdly, mea- sures that rely, predominately, on data from sources exogenous to the port. These three categories are discussed in subsequent sections of this chapter. The Container Port Performance Index 2020 34 2.2.  Measures of Operational or Financial Performance 23. Some of the initial attempts to define and introduce consistent port performance indicators occurred in the early 1970s. In the early 1970s the Secretariat of the United Nations Conference on Trade and Devel- opment (UNCTAD), in response to a formal request, prepared a manual on port statistics (UNCTAD 1971), with the objective of providing guid- ance on better statistical processes for countries and ports. A subse- quent publication, on berth throughput (UNCTAD 1974), included a chapter on port performance indicators, but at that time placed less emphasis on containers per se. In the late 1970s, UNCTAD proposed that signatory countries should adopt and follow certain port perfor- Some of the initial mance indicators (UNCTAD 1987). The proposal’s aim was straightfor- attempts to define ward: to be able to improve performance, countries first needed to and introduce consis- measure performance, beyond the traditional cost accounting mea- tent port performance sures. Without accurate performance indicators, countries also found indicators occurred it difficult to identify operational constraints, define operational perfor- in the early 1970s. mance measures, and plan port development and enhancement needs. 24. The indicators identified and proposed were predominantly related to the operational or financial performance of a port or terminal. In the former case, they were based, generally, on measures of through- put volumes—container movements per crane, container movements per hour, berth utilization, the (total) number of employees in the port (divided by the total number of containers). In the latter case, the financial measures were a move toward activity based costing, to instill an understanding of the vessel and cargo revenues, service revenues, labor costs, and required investment for different berths (or berth groups in the vernacular). Table 2.1 summarizes the indicators proposed by UNCTAD in 1976. These were proposed to be calculated monthly for each berth group servicing a cargo class. A Comparable Assessment of Container Port Performance 35 Table 2.1. Summary of Performance Indicators Financial indicators Operational indicators Tonnage worked Arrival date Berth occupancy revenue per ton of cargo Waiting time Cargo handling revenue per ton of cargo Service time Labor expenditure Turnaround time Capital expenditure per ton of cargo Tonnage per ship Contribution per ton of cargo Fraction of time berthed ships worked Total contribution Number of gangs employed per ship per shift — Tons per ship-hour in port — Tons per gang hours — Fraction of time gangs idle Source: UNCTAD 1976. 25. In 1987, UNCTAD published the Manual on a Uniform System of Port Statistics and Performance Indicators. Produced for the Port Man- agement Associations of East and Southern Africa as well as West and Central Africa, the manual followed a seminar organized by the World Bank in Ghana in April 1986. The Port Management Association of East and Southern Africa made a similar request to the Sub-Saharan Africa Transport Program in 2018, which provides an indication of the prog- ress in introducing a consistent and comparable set of indicators. 26. That same year, UNCTAD partnered with the International Association of Ports and Harbors (IAPH) to produce a monograph on measuring port performance and productivity. Monograph No. 6 was prepared by an external party (De Monie 1987, 2–11), who noted the long history and the large number of studies, reports, and conferences on the topic. De Monie also noted the outcome, generally, had left the majority of parties dissatisfied, if not downright frustrated. The reasons for such an outcome were summarized as the following: firstly, the sheer number of different parameters involved; secondly, the lack of reliable, accu- rate, up-to-date data, collected and collated in a consistent manner; thirdly, the profound influence of local factors on the data collected; and finally, the divergence in data interpretation by different parties. The Container Port Performance Index 2020 36 27. In 2004, UNCTAD developed the Liner Shipping Connectivity Index (LSCI). The LSCI was initially generated on the country level, and since 2018 has also been developed on the port level in collaboration with MDS Trans- modal.1 The index is generated from data on container ship deployment with six components: (i) The number of weekly calls; (ii) the number of competing carriers; (iii) the number of services; (iv) the size of the largest ship that calls in port; (v) the total deployed container carrying capacity, and (vi) the number of other ports connected through direct services. While not directly measuring performance, the LSCI has been shown to be highly correlated with performance and low trade costs. 28. More recently, UNCTAD proposed a port performance scorecard as part of their “TrainForTrade Port Management Program.”2 The main idea was to bring industry and academic perspectives together in terms of port performance in order to share best practices among port managers, particularly from developing countries. While the focus and The main idea was emphasis in the academic research will be discussed in the next sec- to bring industry tion, many of the challenges summarized in De Monie (1987) remain and academic per- relevant today. One could also posit that the intensification of com- spectives together petition between the shipping lines, and the ports themselves, has in terms of port per- led to greater caution and less trust in respect of information sharing formance in order to between public and private stakeholders. The web-based Port Reform share best practices Toolkit (World Bank 2008) included addressed performance indicators among port managers. with the notion of a “port time accounting” system, similar but more detailed, to the early work in UNCTAD’s Monograph No. 6 (De Monie 1987). Kent et al. (2014) utilized a similar approach to measure port per- formance in Central America. But the PORTOPIA (PORTs Observatory for Ports Indicator Analysis) initiative, supported by the European Union and which ended in August 2017, illustrates the ongoing challenge of this type of approach. 29. Since 2018, UNCTAD has also published median time ships spend in port, based on automatic identification system (AIS) data, in collab- oration with MarineTraffic.3 While the reporting is done only on the country level, the information is provided not only for container ships, but for all major shipping markets, including dry and liquid bulk ships. Globally, the data show the time spent in port by container ships, on average, equals only one-third of that spent by dry bulk carriers; how- ever, large differences between countries’ performances do exist. A Comparable Assessment of Container Port Performance 37 2.3.  Measures of Efficiency 30. The literature on efficiency in the port industry emerged about three decades after the construct of technical efficiency was first proposed. The literature on efficiency in the port industry emerged about three decades after Farrell (1957) proposed the concept of “technical effi- ciency.” Farrell’s construct estimates the efficiency of a port calculated by measuring the difference between observed production and theo- retical potential production, the latter defined by the practices of the “better” performing ports in a sample. The quantification of efficiency is linked to the estimation of a frontier that represents a standard against which efficiency is measured. The literature on 31. Two approaches have been developed to estimate the technical effi- efficiency in the port ciency frontier: linear programming techniques, mainly (i) data envel- industry emerged opment analysis (DEA); and (ii) econometric (stochastic) approaches. about three decades Each of the approaches has advantages and disadvantages, and the after Farrell (1957) pro- choice reflects the particular case under study. From a methodological posed the concept of point of view, Cullinane et al. (2006), shows the results obtained using “technical efficiency.” the two approaches are reasonably correlated. 32. Since the 1990s, a proliferation of research on the efficiency of ports has emerged. The first research, in the 1990s, estimated technical effi- ciency of ports in the United Kingdom and Spain. Since 2000, the geo- graphical scope of the studies has been extended to South Asia, Latin America, and Eastern and Southern Africa, along with others. Much of the research supports the unsurprising premise that increased effi- ciency reduces transport costs (see Herrera Dappe et al. 2017; Nordås and Piermartini 2009; and Trujillo et al. 2013). Box 2.1 summarizes one recent study conducted by the World Bank (2018). 33. Some of the results, however, should be interpreted with a note of cau- tion. Despite the apparent sophistication of the statistical and econo- metric approaches, the studies share many of the same challenges as previous attempts. These include differences in the definition and reli- ability of the input statistics (among the questions to address: Does the container handling space include, or exclude container depots outside or adjacent to the port area? Does it include the general cargo berths where overspill containers are handled?), temporal consistency in the data available from different ports and terminals, public sector man- dates to allow more or less storage time, and similar limitations in trust and transparency between terminal operators and port authorities. The Container Port Performance Index 2020 38 Box. 2.1. Port Development and Competition in Eastern and Southern Africa: Prospects and Challenges The current study measures port efficiency using a stochastic frontier production function. The production frontier shows the maximum output quantity that can be obtained with a given com- bination of inputs. The efficiency of a port is calculated by measuring the difference between observed production and theoretical potential production, the latter based on the practices of the “better” performing ports in the sample of similar ports. The logic of using a matched sample of similar ports is that the larger ports will reach levels of technical efficiency that are unattainable by ports of the scale of the Eastern and Southern Africa (ESA) ports. The analysis considered three main input variables: the sum of the length of all container and multi- purpose berths in the port, the total container terminal area of the port, and the combined capacity of the cranes, including ship-to-shore (STS) gantry cranes, and any mobile lifting capacity in the port with a capacity over 15 tons. In addition, several environmental variables were defined to ensure other key contextual factors were reflected in the analysis. These included a dummy variable if there is at least one privately operated terminal, a dummy variable if the port has railway access, and a connectivity index.4 A time trend was included to reflect the effect of technological change. Figure B2.1.1. Average Technical Efficiency among ESA Ports, 2008–17 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Mombasa Djibouti Dar es Maputo Toamasina Cape Durban Port Nacala Port East Beira Salam Town Elizabeth Louis London Globally, the port of Mombasa, based on this data, is the most technically efficient port in the ESA, and ranks as the 43rd most efficient container port in the global sample. Dar es Salaam and Durban follow at 64th and 70th positions respectively. The analysis also reveals the main factors found to drive higher efficiency in container terminal operations in the port are: (i) the presence of at least one specialist terminal operator; (ii) the existence of a rail connection to the port; (iii) the existence of transhipment traffic; (iv) a higher score on the LSCI; and (v) reduced time at berth. Source: World Bank 2018. A Comparable Assessment of Container Port Performance 39 34. Even if the data are comparable, reliable, and complete, interpre- tations will differ within each context. As in one example, a port or terminal under stress might appear an exemplar in terms of the com- parison, as the pressure forces an improvement in the utilization of space, and improvement in operational practices. By contrast, a port under less stress might not face the same pressures, and in some cases encourage higher dwell time for revenue maximization. The picture is obviously nuanced, and the key message is that the various metrics should not be considered alone without carefully considering other indicators within a given context. 2.4.  Measures Using Data from Exogenous Sources 35. The Journal of Commerce (JOC) Port Productivity Index and the jour- ney to the Container Port Performance Index (CPPI) 2020. The year 2009 was one of the worst in terms of losses in the history of container shipping. The financial crisis triggered by the 2008 Lehman Brothers collapse resulted in accumulated losses among container lines reached The year 2009 was one US$20 billion as they dealt with a sharp drop in global demand, low of the worst in terms freight rates and high fuel prices. It was an existential moment for an of losses in the history industry that badly needed to cut costs. of container shipping. 36. In March 2009, at a JOC-organized container shipping conference, a chief executive officer (CEO) of a major shipping line presented the idea of liner shipping companies submitting port call timestamp data for the purpose of performance benchmarking. The rationale included the idea that aggregated port call timestamp data could support initia- tives to improve port call efficiency, resulting in vessels being moved out of port more quickly, allowing them to meet their schedules or slow steam to the next port. Slow steaming would result in reduced fuel use and lower overall fuel costs. The improved efficiency also held the potential to reduce emissions, although decarbonization initiatives did not have the same level of urgency for the industry at that time as they do today. The Container Port Performance Index 2020 40 37. For JOC, the prospect of developing a project that could help drive improvement in container port call efficiency sparked interest and led to the launch of the JOC Port Productivity Project, a precursor of the CPPI. In the beginning, the JOC project focused entirely on berth pro- ductivity—the number of container moves divided by hours at berth (whether gross or net). In the early years, five major carriers provided data monthly to the initiative. The carriers willingly shared data and The carriers willingly were keen to develop and apply the benchmarking to seek out oppor- shared data and were tunities for improving port call processes. In 2012 the project had keen to develop and generated enough traction to warrant a presentation to the Box Club— apply the benchmark- formally known as the International Council of Containership Opera- ing to seek out oppor- tors—at its annual meeting of liner shipping CEOs in New York. The Box tunities for improving Club update spawned more interest and support among major carriers port call processes. resulting in new members providing data on a regular basis. 38. From a technical point of view, a step change in the advancement of the CPPI came after IHS Inc. acquired or became responsible for man- aging several data-rich assets. Over 2008 and 2009, IHS fully acquired the former Lloyd’s Register of Ships, one of the world’s premier ves- sel characteristics datasets. In 2014, IHS Inc. acquired the JOC Group, which publishes the Journal of Commerce. In addition, the IHS data team manages the International Maritime Organization (IMO) vessel numbering scheme as well as the global fleet tonnage assessment for IMO fees. IHS also owns and operates a premium AIS service with both satellite and terrestrial coverage offering both live and historical ship movements data. 39. The mapping of carrier call data to AIS historical vessel movements data provided a key in the development of CPPI, allowing for a new level of quality control due to the capacity to track and verify each individual call. Detailed AIS mapping and geofencing of port and ter- minal zones as well as the capacity to track vessels with high fidelity on approach and within ports also facilitated the creation of productivity metrics based on an exogenous source. At the same time, through more regular dialogue with the carriers, the program made significant progress in standardizing definitions and terms. The 2016 merger of IHS Inc. with leading financial industry data and information provider Markit—to form IHS Markit—led to further improvement of the collec- tion, management, and quality control processes associated with the port performance data. A Comparable Assessment of Container Port Performance 41 40. In autumn 2019 the IHS Markit Port Performance Program5 was pre- sented again to the Box Club, this time in Shanghai. One of the rea- sons the program was added to the Box Club agenda because earlier that year the IMO had identified the potential of port call optimization to contribute toward meeting industry decarbonization targets. The dataset supports identification of gaps and opportunities for improve- ment in container port call processes and can play a role in supporting initiatives aimed at moving the industry toward just-in-time (JIT) vessel arrivals and the associated benefits of that in terms of reduced fuel consumption and reduced emissions. 41. After the 2019 Box Club meeting, the number of carriers supporting Currently, the Port the program quickly rose to 10, which combined control over 76 per- Performance Program cent of the global container fleet capacity. Currently, the Port Perfor- accounts for more mance Program accounts for more than 180,000 port calls and 240 mil- than 180,000 port calls lion container moves per year. The performance dataset covers more and 240 million con- than 1,000 terminals in over 500 container ports. This critical mass, tainer moves per year. combined with the high-quality data collection, verification, manage- ment and distribution put in place over the past decade has created a valuable resource that for the first time allows us to credibly and objec- tively measure and compare ocean-side container port and terminal performance on global scale. 42. IHS Markit and the World Bank are drawing on this resource to part- ner in the development and publication of the CPPI. A more detailed discussion of the data behind the construction of the CPPI, the logic, and different approaches employed, are summarized in chapter 3, while chapter 4 presents the results of the inaugural CPPI 2020. 1. UNCTAD offers country- and port-level LSCI data, along with bilateral (country-to-country) LSCI data NOTES online: http://stats.unctad.org/maritime/, while MDS Transmodal offers LSCI for the top 100 ports: https://www.portlsci.com. 2. See https://tft.unctad.org/port-management/ for more information on UNCTAD’s TrainForTrade’s Port Management Program. 3. Explore the MarineTraffic live shipping map online: https://marinetraffic.com. 4. See the Review of Maritime Transport 2020 (UNCTAD 2020) to learn more about the latest iteration of the connectivity index (LSCI) used in this 2018 study. 5. For more information on the Port Performance Program, see: https://ihsmarkit.com/products/port-performance.html. The Container Port Performance Index 2020 42 REFERENCES Cullinane, Kevin, and Dong-wook Song. 2006. “Estimating the Relative Efficiency of European Container Ports: A Stochastic Frontier Analysis.” Research in Transportation Economics 16: 85–115. https://doi.org/10.1016/S0739-8859(06)16005-9. De Monie, G. 1987. Monograph No. 6: Measuring and Evaluating Port Performance and Productivity. UNCTAD Monographs on Port Management Series. Geneva: UNCTAD/IAPH. https://unctad.org/system/files/official-document/ship4946_en.pdf. Farrell, M. J. 1957. “The Measurement of Productive Efficiency.” Journal of the Royal Statistical Society Serie A, Part III 120 (3): 253–67. http://dx.doi.org/10.2307/2343100. Herrera Dappe, Matias, Charl Jooste, and Ancor Suárez-Alemán. 2017. “How Does Port Efficiency Affect Maritime Trans- port Costs and Trade? Evidence from Indian and Western Pacific Ocean Countries.” Policy Research Working Paper 8204, World Bank, Washington, D.C. http://documents.worldbank.org/curated/en/388141506343576100/How- does-port-efficiency-affect-maritime-transport-costs-and-trade-evidence-from-Indian-and-western-Pacific-Ocean- countries. Kent, P. E., G. Ayzanoa, and A. Ashar. 2014. “How Fit Are Central America’s Ports? An Exercise in Measuring Port Perfor- mance.” Paper presented at the International Association of Maritime Economists Conference, Norfolk, Virginia, July 15–18, 2014. Nordås, Hildegunn Kyvik, and Roberta Piermartini. 2009. “Infrastructure and Trade.” WTO Staff Working Paper ERSD 2004-4, World Trade Organization, Economic Research and Statistics Division, Geneva. https://ideas.repec.org/p/zbw/wtowps/ersd200404.html. Trujillo, Lourdes, María Manuela González, and Juan Luis Jiménez. 2013. “An Overview on the Reform Process of African Ports.” Utilities Policy 25:12–22. https://doi.org/10.1016/j.jup.2013.01.002. UNCTAD (United Nations Conference on Trade and Development). 1971. Port Statistics: Selection, Collection and Presentation of Port Information and Statistics. Geneva: UNCTAD. https://unctad.org/system/files/official-document/tdbc479rev1_en.pdf. ——. 1974. Berth Throughput: Systematic Measures to Improve General Cargo Operations. Geneva: UNCTAD. ——. 1976. Port Performance Indicators. Geneva: UNCTAD. https://unctad.org/system/files/official-document/tdbc4d131sup1rev1_en.pdf. ——. 1987. Manual on a Uniform System of Port Statistics and Performance Indicators. Geneva: UNCTAD. https://unctad.org/system/files/official-document/shipd185rev2_en.pdf. ——. 2020. Review of Maritime Transport 2020. Geneva: UNCTAD. https://unctad.org/webflyer/ review-maritime-transport-2020. World Bank. 2008. “Port Reform Toolkit: Second Edition.” Online Toolkit, PPIAF/World Bank, Washington D.C. https://ppp.worldbank.org/public-private-partnership/library/port-reform-toolkit-ppiaf-world-bank-2nd-edition. World Bank. 2018. “Port Development and Competition in East and Southern Africa: Prospects and Challenges.” World Bank, Washington, D.C. https://openknowledge.worldbank.org/handle/10986/31897. A Comparable Assessment of Container Port Performance 43 3.  The Approach and Methodology 3.1. Introduction 43. Container (liner) shipping services are generally highly structured, ser- vice rotations. These services are typically set up with weekly departure frequencies, fixed sequence of port calls, and standard pro forma day- and time-specific berthing windows. Once a service has been defined or adjusted, it will usually remain intact for many months, or even years. The berthing windows are pre-agreed with the terminal and port operators, usually based on a slightly higher than expected average quantity of container exchange moves, and ideally with modest buffers included in the sea legs between ports. The Container Port Performance Index 2020 44 44. The clear advantages of a highly structured service model include: (i) shippers can make long-term supply decisions, and (ii) ports and terminals can schedule and balance their resources to meet expected With a well-planned demand. With a well-planned and well-executed pro forma schedule, and well-executed higher levels of reliability and predictability can be achieved, which is pro forma schedule, advantageous for more effective supply chain operations and planning higher levels of reli- as containerships will spend around 15 to 20 percent of their total full ability and predictabil- rotation time in ports, with the balance being spent at sea. As noted ity can be achieved. previously in the report, reduced port time can allow ship operators to reduce vessel speed between port calls, thereby conserving fuel, reduc- ing emissions, and lowering costs in the process. 45. Conversely, for every unplanned additional hour in port or at anchor- age, the ships will need to increase speed to maintain the sched- ule, resulting in increased fuel consumption, increased costs, and increased emissions. In extreme cases, ships that fall many hours behind their pro forma schedule will start to arrive at ports outside of their agreed windows, causing berth availability challenges for ports and terminals, particularly those with high berth utilization rates. This in turn can cause delays to shipments and disruption to supply chains. In extreme cases, A service recovery can involve significantly higher sailing speeds, ships that fall many increasing fuel consumption, emissions and costs, or the omission of a hours behind their pro forma schedule port or ports from the standard rotation, disrupting supply chains and will start to arrive at often resulting in additional contingency costs. ports outside of their agreed windows, caus- 46. Because time is valuable for stakeholders, a logical step is to mea- ing berth availability sure port performance based upon the total amount of time ships are challenges for ports required to spend in port. To facilitate this measurement, the Container and terminals, partic- Port Performance Index (CPPI) 2020 has been developed based on total ularly those with high port time as explained in subsequent report sections. This first CPPI berth utilization rates. iteration utilizes data through June 30, 2020. Similarly, the next version (CPPI 2021) will utilize data through June 30, 2021. The construction of the CPPI 2020 has employed two different approaches, an administra- tive approach and a statistical approach, to ensure the resulting rank- ing of container port performance reflects as closely as possible actual port performance, while also being statistically robust. The approaches are discussed in this chapter, with further detail on the statistical meth- odology provided in appendix B. The results are presented in chapter 4. A Comparable Assessment of Container Port Performance 45 3.2.  The Anatomy of a Port Call 47. Every container port call can be broken down into six distinct steps. Figure 3.1 illustrates these individual steps. The term total port hours is defined as the total elapsed time between when a ship reaches a port (either port limits, pilot station, or anchorage zone) to when it departs from the berth after completing its cargo exchange. The time expended from berth departure (all lines up) to the departure from the port limits is deliberately excluded. The justification for the exclusion is any per- formance loss that pertains to departure delays, such as (i) pilot or tug availability; (ii) readiness of the mooring gang; (iii) channel access and water depths; and (iv) ship readiness, will be incurred while the ship is still alongside. Additional time resulting from these causes will there- fore be captured when the clock stops at berth departure. Figure 3.1. The Anatomy of a Port Call Source: Figure graphic provided by IHS Markit, Marine & Trade division The Container Port Performance Index 2020 46 48. Ships may spend additional time in a port after the departure from a berth; however, the time associated with these additional activities is excluded from the CPPI. Ships might dwell within a port’s limits for reasons that include bunkering, repairs, or simply waiting in safe areas if unable to berth on earliest arrival at the next port. Except for bunker- Ships may spend ing not being performed simultaneously with cargo operations, these additional time in a causes of additional port time do not necessarily reflect inefficiency or port after the depar- poor performance of the port and are thus excluded from the CPPI. ture from a berth. They might dwell 49. Conversely, ships occasionally might not sail immediately upon com- within a port’s lim- its for bunkering, pletion of cargo operations because they are not operationally ready, repairs, or simply have not been released by local authorities, or remain alongside to waiting in safe areas. complete repair works or bukering. With the exception of authority clearance delays, none of the other causes of additional port time are reflective of port inefficiency per se. However, the available data are insufficiently granular to see root causes of delays by their nature. The Port Performance Program presumes the percentage of ships idling alongside after completion of cargo operations for reasons unrelated to port performance is modest, and thus continued inclusion will have no significant effect on the CPPI results. 50. The other four components of the port call can logically be grouped into two distinct blocks of time. The first comprises elapsed time between arrival port limits and all lines fast (steps 1 to 2 in figure 3.1); the second comprises time elapsed between all lines fast and all lines up (steps 2 to 5, also commonly referred to as on-berth time or berth Waiting time, defined hours). The logic behind this division contends there will always be time as the period between arrival port lim- consumed between steps 2 and 5, while the bulk of time between steps its, which includes 1 and 2—excluding actual sailing time—is waiting time that could be the anchorage zone, reduced to zero. and all lines fast is generally regarded 51. The issue of waiting time in the measurement of port performance. as pure waste. Waiting time, defined as the period between arrival port limits, which includes the anchorage zone, and all lines fast is generally regarded as pure waste. As such, in the construction of the CPPI, one possibility was to apply a penalty to waiting time. The decision was taken not to, as the introduction of a penalty of this type could be seen as a normative judgment, which would be inconsistent with the program’s overarching objective of producing an objective quantitative index. A Comparable Assessment of Container Port Performance 47 52. On a related note, the CPPI team deliberated in whether to apply a “discount” to waiting time for the smallest segment of ships. This question arose because ships with less than 1,500 twenty-foot equiva- lent units (TEUs) of nominal capacity spend notably more time waiting at hub ports1 than all other ship size segments (see figure 3.2 and table 3.1, which display “arrival” times; however, this disparity is not present at all hub ports). The disparity could exist because larger ships gener- ally enjoy a higher priority for berthing. It could also be partly explained due to the practice of feeder ships waiting for cargo to become avail- able, close to where they expect that demand to materialize. Neither circumstance is linked to port performance. Figure 3.2. Average Arrival Times at Hub Ports per Ship Size Range 5.0 4.6 4.5 4.0 3.5 Hours 3.5 3.3 3.0 3.0 3.0 2.9 2.7 2.5 2.0 <1,500 >1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 Overall Ship size range (nominal TEU) Source: Original calculations for this publication, based on CPPI 2020 data. The Container Port Performance Index 2020 48 Table 3.1. Disparity in Arrival Times at Hub Ports Port Ship size less Ship size greater than 1,500 TEU than 1,500 TEU (hours) (hours) Cartagena (Colombia) 9.1 2.2 Alexandria (Egypt) 10.4 3.8 Jeddah 9.5 3.0 Valencia 9.7 3.2 Damietta 9.7 4.7 Manzanillo (Mexico) 8.8 4.2 Rotterdam 8.3 4.4 Hamburg 7.8 4.1 Piraeus 6.2 3.2 Balboa 5.2 2.5 Bremerhaven 5.4 3.2 Port Louis 4.8 2.7 Source: Original calculations for this publication, based on CPPI 2020 data. Note: TEU = twenty-foot equivalent unit 53. Because the performance data do not indicate whether waiting time is voluntary or forced, finding a suitable level at which to discount wait- ing time in this scenario can be challenging. Port calls of ships less than 1,500 TEU capacity comprise just 10 percent of total calls in the CPPI, making a minimal impact to the overall CPPI of the disparity in waiting times between ships with less than 1,500 TEU of nominal capacity and other segments. The same applies with respect to this issue being a normative judgment inconsistent with an objective quantitative index. A Comparable Assessment of Container Port Performance 49 3.3.  The Data Underpinning the CPPI 54. The data used to compile the CPPI are taken from the IHS Markit port performance program database. Started in 2009, the port performance program goal is to drive efficiency improvements in container port operations and support programs to optimize port calls. The program currently comprises 10 of the world’s largest liner shipping compa- nies, which collectively operate 76 percent of global fleet capacity. This Started in 2009, the unique and highly comprehensive dataset covers 502 ports and 1,014 port performance pro- terminals in 137 countries, with calls by 3,860 individual vessels. On an gram goal is to drive annual basis, the database comprises more than 180,000 port calls. efficiency improve- ments in container 55. Because the source data are provided by the larger container ship port operations and operators, performance results are slightly skewed toward the larger support programs to container ship size classes. However, many program participants have optimize port calls. sister or subsidiary companies that specialize in intraregional or feeder services or they might operate such services under their main brands. Thus, call data from these services are included in the dataset and, by extension, the CPPI. 56. In 2015 and 2016, participating members agreed on standard metric definitions. This agreement ensured comparisons could be drawn from consistent empirical data. Member shipping lines provide the port call data to IHS Markit on a monthly basis. The data undergo a thorough checking and validation process, including removal of records falling outside well-defined data quality boundaries that usually exist as a The port call data result of data input errors. The data are validated against IHS Markit’s provided by member geodatasets to ensure full harmonization of location names and vessel. shipping lines, undergo The received data are matched with proprietary historical AIS (auto- a thorough checking matic identification system) port timestamps for validation, and cre- and validation process. ation of additional performance metrics. 57. The AIS matching process involves taking the berth arrival date and time provided by the member shipping lines and searching back through the historical AIS data. The objective is to determine the ear- liest arrival date and time for a port or anchorage zone mapped to the destination port. The search backwards stops at berth arrival minus 144 hours (six days). Searching beyond the 144-hour boundary is not feasible as it would likely result in capturing port arrival dates and times for port calls from previous weeks. These data would consist mainly of feeder ships running on weekly frequencies or mainline ships where a sailing was blanked or voided, or ships being in repair yards immedi- ately prior to a port call. The Container Port Performance Index 2020 50 58. The success rate of the AIS matching process ranks consistently around 90 percent. The failed match quantity of 10 percent normally relates to locations with a shortage of land-based terrestrial receivers, which requires higher dependence on satellite tracking—a less reliable tracking method in busy port areas. A review of the port-level data indi- cates those ports where AIS timestamp match ratios are low (see table 3.2). Port hours cannot be calculated without a valid port arrival date and time. These included port calls at a further 75 ports,2 amounting to a collective 1,043 port calls, though these ports do not feature in the CPPI 2020 iteration. Table 3.2. Ports with Low Levels of Matched AIS Data Port name Matched Port name Matched (percent) (percent) Algiers 37 Latakia 30 Alicante 20 Magdalla 15 Aqaba 35 Muhammad Bin Qasim 20 Bar 9 Mundra 30 Chennai 10 Nassau 22 Chu Lai 30 Pipavav 18 Corinto 29 Ploce 0 Damietta 39 Prachuap Port 0 Fort-De-France 35 Puerto Bolivar (Colombia) 0 Freetown 19 Qingdao 12 Kamarajar 3 Rades 7 Kattupalli 7 San Carlos (Philippines) 0 Kribi Deep Sea Port 30 Subic Bay 13 Krishnapatnam 5 Source: Original calculations for this publication, based on CPPI 2020 data. A Comparable Assessment of Container Port Performance 51 3.4.  The Construction of the CPPI 59. For a port to qualify for inclusion in the CPPI it must have registered at least ten valid port calls where port hours can be calculated within each six-month period. Out of the 502 ports where IHS Markit receives port call information, a total of 351 ports (765 terminals and 2,877 individual vessels) are included in the CPPI 2020. Over this time period, 67,798 distinct port calls were recorded in the data, representing more than 50% percent of the world’s total container ship port calls. Future iterations of the CPPI will look to include data for additional, often smaller, ports that have significant issues with AIS coverage. 60. The CPPI is based on the overarching metric of port hours per ship call. However, to account for significant differences in ship calls determined by: (i) greater or lesser workloads; and (ii) smaller or larger capacity ships, calls are analyzed in ten narrow call size groups and five ship size groups that generally reflect the types of ships deployed on specific trades and services (figure 3.3). Figure 3.3. The Structure of the CPPI Source: Original calculations for this publication, based on CPPI 2020 data. Note: TEU = twenty-foot equivalent unit The Container Port Performance Index 2020 52 61. An average of approximately 75 percent of total port time is con- sumed by container handling operations. As figure 3.4 shows, the ratio of time spent on handling operations increases as ship size increases, and is driven by larger call sizes. Call size does not affect the amount of time spent on arrival, or start and finish processes. Rather, these processes are influenced by ship size. Mooring operations and lashing completion generally take longer on larger ships, although this delay is somewhat offset by larger ships potentially enjoying a slightly higher priority and assignment of resources. Figure 3.4. In-Port Time per Port Stay Step or Process3 Other port hours Other berth hours Operating hours 100 11.3% 10.4% 8.5% 8.5% 90 14.9% 19.5% 8.7% 12.7% 10.5% 80 13.5% 16.9% 15.8% 70 Percent 60 50 40 79.1% 82.8% 74.6% 71.6% 76.0% 30 64.7% 20 10 0 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 Overall Ship size range (nominal TEU) Source: Original calculations for this publication, based on CPPI 2020 data. Note: TEU = twenty-foot equivalent unit A Comparable Assessment of Container Port Performance 53 62. A correlation of close to 60 percent is observed between call size and crane intensity using a logarithmic trend line. As figure 3.5 illustrates, crane intensity increases rapidly through smaller call sizes and starts to peak and flatten when call sizes reach 4,000 to 6,000 moves per call. With a maximum observed crane intensity of around nine cranes per ship, it is interesting to note that for calls exceeding 11,000 moves, crane intensity is always below 6.0. This consistent result reflects a global representation; most of the call sizes in excess of 11,000 moves occur on North America’s west coast where ports have multiple termi- nals, each with a limited quantity of cranes (crane density), and where the deployment of additional cranes is generally considered to be cost prohibitive. Above and beyond crane density lie factors relating to how close together adjacent cranes can operate. Additional constraints from the stowage plan can also have an impact if an excessively intense area of work is concentrated in one part of the ship. Figure 3.5. Regression Analysis: Crane Intensity4 versus Call Size 10.0 9.0 Crane intensity (quantity of cranes used per port call) 8.0 7.0 y = 1.1974In(x) – 5.3309 ... R2 = 0.5993 6.0 5.0 4.0 3.0 2.0 1.0 0 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 Call size (moves per port call) Source: Original calculations for this publication, based on CPPI 2020 data. The Container Port Performance Index 2020 54 63. Average call size increases as ship size increases, but not proportion- ately with ship size (see figure 3.6). This result is partially influenced by the deployment trends and dynamics of the liner shipping networks, and the capacity and volume of a particular port. For example, a ship of 20,000 TEU capacity will likely be deployed on the Asia–North Europe trade lane and make more than 12 port calls per rotation. On the other hand, a ship of 14,000 TEU deployed on the transpacific route would likely serve just eight ports in a full rotation. If all rotations had an equal quantity of port calls, then call size and ship size would trend propor- tionately, assuming that (i) a port can physically handle vessels of an increased size; and (ii) availability of sufficient containers to be loaded and unloaded to justify a stop by a larger-scale vessel. Figure 3.6. Relationship between Global Ship Size Index and Call Size Index 16.0 CS Index SS Index 14.0 12.0 10.0 Index 8.0 6.0 4.0 2.0 – <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 Ship size range (nominal TEU) Source: Original calculations for this publication, based on CPPI 2020 data. A Comparable Assessment of Container Port Performance 55 64. The gross crane productivity (or GCMPH) by ship size group displayed in figure 3.7 represents a global average across all ports. The average gross crane productivity performance for ship size groups demon- strates a limited variation in ships with capacities of 5,001 to 8,500 TEU and larger. Average gross crane productivity is lower for the two smaller ship size groups, albeit by a maximum of around 14 percent between the groups comprising ships under 1,500 TEU and above 13,500 TEU. A different result might have been expected because height and out-trolley distances are greater on larger ships. However, the additional cycle time necessitated by the greater distances is likely compensated for by a larger quantity of moves per bay and more moves per hatch-cover move and gantry position. 65. Smaller ships also tend to suffer more from trim and list in port and generally deploy poorer lashing systems. This would further account for the slightly lower call size performance observed in ships with capacities of 5,000 TEU and under. Analysis of crane intensity and crane productivity shows that crane intensity has a major impact on operat- ing time, and by extension 75 percent of total port time. The evidence shows, therefore, that in order for the CPPI to be realistic, both ship and call size segments should be reflected in the index construction. Figure 3.7. Gross Crane Productivity or Gross Crane Moves per Hour (GCMPH) 26.0 25.0 24.0 GCMPH 23.0 22.0 21.0 20.0 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 Ship size range (nominal TEU) Source: Original calculations for this publication, based on CPPI 2020 data. The Container Port Performance Index 2020 56 66. When comparing port performance and efficiency, taking note of and accounting for the extent of this impact is essential. This can be done in two ways: by applying a weighting by call size, or by examin- ing performance within call size groups narrow enough to effectively neutralize the influence of call size and crane intensity. After extensively testing both methods, findings indicate the former can generate some volatile results when applied across a broad range of ports and ship sizes, and therefore the index development team chose the latter as being the more reliable for the CPPI methodology. 67. A progressive increase trend in crane intensity emerges as the data move through the call size groups. As shown in figure 3.8, this trend peaks and flattens once CPPI data pass the 6,000 moves per call range, leading to the presumption that crane intensity does not have a major influence when call size exceeds 6,000 moves per call. Figure 3.8. Average Crane Intensity per Defined Call Size Range 26.0 25.0 24.0 GCMPH 23.0 22.0 21.0 20.0 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 Ship size range (nominal TEU) Source: Original calculations for this publication, based on CPPI 2020 data. A Comparable Assessment of Container Port Performance 57 68. In order to assess the sensitivity within each call size group across all 351 qualifying ports, the CPPI considers the median call size between all ports within a call size group, with a tolerance range of 10 percent above and below the median created (see table 3.3). In the five call size groups, from 1,001–1,500 to 3,001–4.000 inclusive, most ports have an average call size well within this tolerance range. When one goes beyond the threshold of 4,000 moves per call, call size has much less of an impact on crane intensity, since the number of deployable cranes is limited by the overall number of available or deployed cranes and stow- age splits. The quantity of ports with an average call size within the tol- erance range in the three smallest call size groups is not as high as the quantity in the five call size groups from 1,001–1,500 to 3,001–4,000. However, for those ports with an average call size above the tolerance range, it would be possible to increase crane intensity to match the higher call sizes, and therefore the CPPI results indicate these ports are not being penalized unfairly. Table 3.3. Call Size Sensitivity Call size sensitivity Call Size group (number of moves) <250 251–500 501– 1,001– 1,501– 2,001– 2,501– 3,001– 4,001– >6,000 1,000 1,500 2,000 2,500 3,000 4,000 6,000 Average 180 378 730 1,231 1,730 2,232 2,734 3,435 4,721 7,712 Median 189 379 728 1,216 1,714 2,226 2,728 3,405 4,668 7,174 Range low 171 341 655 1,095 1,543 2,003 2,455 3,064 4,201 6,457 Range high 208 417 801 1,338 1,886 2,448 3,000 3,745 5,135 7,892 Total ports 270 324 318 260 218 173 135 116 81 41 Within range 141 230 226 239 212 173 135 111 64 25 Percentage in range 52.2% 71.0% 71.1% 91.9% 97.2% 100.0% 100.0% 95.7% 79.0% 61.0% Source: Original calculations for this publication, based on CPPI 2020 data. The Container Port Performance Index 2020 58 69. A practical consideration to keep in mind when looking at the data is the need to keep to a workable number of call size groups. A work- able number of call size groups ensures the data are not overly diluted; therefore, the CPPI does not include additional call size groups for below 1,000 moves per call. After examining multiple options, the CPPI team found an efficiency comparison based on these ten defined call size groups to be both objective and relevant, while also neutralizing the influence of call size and crane intensity. 70. In addition, the application of ship size groups is less important than One objective of the call size groups, particularly once the call data is already split into ten CPPI is to highlight— call size groups. However, one objective of the CPPI is to highlight— through compari- through comparison—the performance gaps, which in turn reveal son—the performance opportunities to save fuel and reduce emissions. The analysis should gaps, which in turn therefore account for the fact that larger ships consume more fuel, and reveal opportuni- thus present a greater potential opportunity to save fuel and reduce ties to save fuel and emissions. reduce emissions. 71. For this reason, the CPPI methodology includes five different ship size groups. Five is considered the optimum number of ship size groups to account for larger ships using more fuel without overly diluting the data (see table 3.4). Each group is defined by clustering similar classes of container ships together and, to the greatest extent possible, the trade lanes where they would generally be deployed. The typology is based on nominal TEU capacity, but it would be equally effective to cat- egorize based on gross registered tonnage or length overall. 72. Some 43 percent of all calls in the database were made by ships with capacity of between 1,500 and 5,000 TEU. The other two midrange ship size bands are also well represented. As expected, the CPPI shows a lower concentration of calls in the smallest and largest ship size bands. Nevertheless, the database holds a sufficiently high quantity of data to produce representative benchmarking in the smallest and largest ship size bands. A Comparable Assessment of Container Port Performance 59 Table 3.4. Ship Size Groups Nominal TEU Description capacity range Less than Comprising the smallest sizes of ships, this group is almost exclusively feeder ves-sels, which often connect small 1,500 TEU outlying ports with regional hub ports. Some in-traregional services will also have ships in this size range. 1,500 to 5,000 TEU A vast quantity of these classic Panamax ships is deployed on intraregional trades. They are found on Intra-Asian trades and north–south trades to and from Africa, Latin America, and Oceania, as well as on transatlantic services. 5,000 to 8,500 TEU Vessels within this size group are mainly deployed on the north–south trade-lanes. Vessel cascading and improv- ing port capabilities has allowed them to start emerging as stock vessels for Africa and Latin America trades as well as Oceania, with some presence on transatlantic and Asia–Middle East trades as well. 8,500 to 13,500 TEU These Neo-Panamax vessels are largely deployed on east–west trades, particu-larly transpacific, both to North America’s west coast as well as via either the Panama or Suez canals to North America’s east coast. They also fea- ture on Asia–Middle East trades, with some deployed on Asia–Mediterranean rotations. Greater than These ultra-large container ships (ULCS) are mainly deployed on Asia–Europe (serving both northern Europe and 13,500 TEU the Mediterranean) and the Asia–USA trades, especially on transpacific services calling at North America’s west coast ports. Source: World Bank analysis. Note: TEU = twenty-foot equivalent unit. 73. Of the potential 50 (5 × 10) ship and call size categories, seven of the categories have insufficient data for meaningful analysis (as indi- cated by the gray cells in table 3.5); therefore, these categories have been omitted from the CPPI. Some of the other call size bands within ship size ranges contain limited data; however, the data are considered sufficient for inclusion and analysis. Data for each ship size group are extracted from the full dataset, and then broken down into average hours consumed within each call size group at each port. Total port hours are broken down into arrival hours—defined as elapsed time from arrival port limits to all lines fast—and berth hours—defined as elapsed time from all lines fast to all lines released. With 43 potential ship and call size combinations per qualifying port, many combinations will naturally occur where no actual data exists. The Container Port Performance Index 2020 60 Table 3.5. Call Size and Ship Size Combinations: Percentage of Calls Ship size group Call Size group (number of moves) (TEU) <250 251–500 501– 1,001– 1,501– 2,001– 2,501– 3,001– 4,001– >6,000 1,000 1,500 2,000 2,500 3,000 4,000 6,000 1) <1,500 27.0% 39.6% 28.5% 4.9% 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% 2) 1,501–5,000 9.8% 24.8% 36.1% 17.1% 7.7% 2.7% 1.0% 0.7% 0.2% 0.0% 3) 5,001–8,500 2.6% 9.7% 27.4% 23.2% 15.6% 9.5% 5.2% 4.5% 1.9% 0.3% 4) 8,501–13,500 1.3% 6.7% 19.1% 18.2% 16.3% 11.9% 9.0% 10.0% 5.2% 2.1% 5) >13,500 0.2% 1.3% 5.5% 7.9% 10.5% 11.6% 11.2% 22.1% 22.4% 7.5% Source: Original calculations for this publication, based on CPPI 2020 data. 74. In preparing the CPPI and its subsequent rankings, the objective has been to reflect as closely as possible actual port performance, while also remaining statistically robust. With respect to the largest ports— the top 100 ports by annual move count—the CPPI has been able to work with ample, real empirical data present in each of the 43 distinct ship size and call size categories. However, for smaller ports many cat- egories have been populated with little to no data, particularly those with only a few hundred calls in total. If these unpopulated categories are ignored, the performance appraisal would be undertaken on dif- ferent quantities of categories, which is likely to unduly disadvantage smaller ports that might well be quite efficient despite their modest size and throughput. Imputing missing values: The administrative approach 75. The CPPI addresses this undeserved handicap with two different approaches: An administrative approach and a statistical approach. In the case of the former, the approach involved assigning values to empty categories based upon data available when a port has registered a data point within a specific ship size range. A Comparable Assessment of Container Port Performance 61 Table 3.6. Quantity of Ports Included per Ship Size Group Ship size range Number of ports Base call size (nominal TEU) included (number of moves) Less than 1,500 219 251–500 1,500–5,000 331 501–1000 5,000–8,500 213 1001–1500 8,500–13,500 162 1501–2000 More than 13,500 99 2001–2500 Source: Original calculations for this publication, based on CPPI 2020 data. 76. For each ship size group, the call size group that has the largest quan- tity of data representation is selected (see table 3.6). Ideally, this is a midrange call size group because the lowest and highest demonstrate some uniqueness. In cases where no actual data are available for the base call size group, the next highest group is examined to find an actual dataset. If none is found, then the approach involves looking at the immediately lower call size band. At the end of this exercise, every port has a value present for the base call size group. 77. From the base call size group, moving left toward the lowest group and right toward the highest group, in groups where no value exists, a value is determined on a pro-rate basis given the adjacent call size group value (actual data or imputed). The rationale is that if within one call size group a port had either higher or lower hours for each arrival and berth hours than the average, it is most probable that would be the case in the adjacent call size group. The arrival and berth hours are imputed individually of each other. 78. Table 3.7 contains an illustrative example. In this case, Port A had a higher quantity of hours in the base call size group than the group average. The index assumes that would also have been the case had the port registered actual calls in the 501 to 1,000 and 1,501 to 2,000 call size groups. The opposite is true for Port B, which achieved a lower quantity of hours in the base call size group. The Container Port Performance Index 2020 62 Table 3.7. An Example of Imputing Missing Values5 Port Call size group (hours) 501–1,000 moves 1,001–1,500 moves 1,501–2,000 moves Port A 10.8 12.0 14.4 Port B 7.2 8.0 9.6 Group average 9.0 10.0 12.0 Factor multiplier 0.9 Base 1.2 Source: Original calculations for this publication, based on CPPI 2020 data. 79. The inherent risk with this approach is that poor or good performance within just one group will cascade across all call size groups. It also assumes a port’s ability to add cranes to larger call size groups, which might not be true in all cases. On the other hand, it would be illogical to blindly assume any port would simply achieve the average of the entire group or, possibly worse, to assume that a port performing below average in one call size group would miraculously perform much better than average in others where it did not record any actual calls. Imputing missing values: The statistical approach 80. A slightly more sophisticated approach is using a likelihood-based method to impute those missing values. In respect of the current dataset, the expectation–maximization (EM) algorithm can be utilized to provide a maximum likelihood estimator for each missing value. This approach relies on two critical assumptions: The first assumption is that the missing values are random, that is, not due to some bias in the sample selection; and the second assumption is that all variables under consideration are normally distributed. Within the context of the dataset, these are considered realistic assumptions. EM then computes the maximum likelihood estimator for the mean and variance of the normal distribution given the observed data. Knowing the distribution that generates the missing data, one can then sample from it to impute the missing values. The “sampling” is repeated a number of times to remove the risk of outliers, and negative numbers are set to zero (as berth time can never be negative).6 A Comparable Assessment of Container Port Performance 63 Constructing the index: The administrative approach 81. Aggregating arrival and berth hours into total port hours. Previously, the report indicated a case could be made for penalizing waiting time, which can be regarded as pure waste. However, as expressed earlier this would be a normative judgment. Accordingly, in the CPPI 2020, both arrival and berth hours are weighted as 1.0 and the two time seg- ments are summed to form total port hours. 82. Appraising port hours performance. Average port hours are naturally Average port hours are higher in the larger than smaller call size groups. This can magnify the naturally higher in the difference in hours between a subject port and the average port hours larger than smaller of the overall group. Thus, appraising on the difference between a port’s call size groups. This average hours and average hours of the group may skew the scoring can magnify the differ- unduly toward the larger call size calls. Because larger call size group ence in hours between have far fewer calls than smaller call size groups, this also needs to be a subject port and the reflected in the construction of the CPPI to retain maximum objectivity. average port hours of the overall group. 83. The method applied to each call size group individually is that the port’s average port hours is compared with the group average as a percentage or rather a ratio. The result of that comparison is weighted by the ratio of port calls in each call size group (within each ship size group) for the entire group of ports (table 3.3, earlier in the chapter, provided the ratios).7 Table 3.8 illustrates how the method is applied to one ship size group (5,001 to 8,500 TEU) and one base call size group (1,001 to 1,500 moves). Table 3.8. Port Hours Performance Appraisal Port Port hours Score Call size group weight Result Example port 19.72 1.152 0.232 0.2672 Group average 22.71 Source: Original calculations for this publication, based on CPPI 2020 data. The Container Port Performance Index 2020 64 84. In this illustrative example, the subject port used 15.2 percent fewer hours than the average of the entire group (19.72 hours versus 22.71 percent). The result is a (ratio) score of 1.152. Since 23.2 percent of all port calls in the 5,001–8,500 TEU ship size group were in the call size group of 1,001–1,500 moves, the 1.152 score is multiplied by ratio 0.232 for an overall result of 0.2672 “points.” Within our tables, we have then multiplied this by 100 to achieve an easier to read number (26.72). The better the performance versus the average of the group, the higher (or lower – negative) the overall result will be. 85. Aggregation to a score and rank per ship size group. The “result” achieved per port within each of the ten call size groups are then added together to calculate a score within the overall ship size group (it is five and eight groups rather than ten groups in the case of the two smaller ship size groups, respectively). Based upon these scores a subranking is performed within each ship size group that can then be reviewed in the final CPPI rankings. 86. However, the imputation method might unfairly appraise some ports that recorded data only within a few call size groups. If, for example, the performance in a few call size groups was worse than the average for all ports within the ship size group, this would be prorated to all call size groups. This requires a judgment, as the alternative of ignoring call size groups without actual data, effectively resulting in a zero score for those groups, would not necessarily result in a better outcome. In the latter case, ports with limited call size diversity would not be credited with positive scores in each and every call size group, which ports with a greater diversity of call sizes are likely to achieve. Aggregating all ship size groups 87. No allowance was made for ports that did not handle ships within specific ship size groups during the period under consideration. Earlier in the chapter, table 3.6 presented the quantity of ports being included per ship size group. Allowance was not given primarily because many of the smaller ports are not capable of handling some of the larger ship sizes and so would in effect be awarded positive (or negative) results for scenarios that are physically impossible. The omission of scores within some ship size groups would only be an issue if attempting to com- pare the performance of major mainline ports with those of far smaller ports; however, this comparison is neither objective nor valuable. A Comparable Assessment of Container Port Performance 65 88. This factor does not contribute significantly to the comparison between similarly sized ports. In aggregating the scores from the various ship size groups into the overall CPPI in the administrative approach, a factor was built, in the form of an index (see table 3.9), to differentiate the importance and significance of better performance on larger ships over smaller ones, based on the relative fuel consumption between different ship sizes. For each ship size group, a typical mid- range example ship was selected. Based upon the expected deploy- ment of such ships, the index defines and weights a range of sea legs, using a typical pro forma service speed, and considering the impact on fuel consumption that one hour longer (or shorter) in port would be likely to yield. Table 3.9. Assumptions to Determine a Fuel Consumption Index Nominal TEU Expected deployment Sea leg Weight Index capacity range (percent) value Less than 1,500 TEU Feeders Singapore–Surabaya 25 Intraregional Rotterdam–Dublin 25 0.46 Kingston–Port-Au-Prince 25 Busan–Qingdao 25 1,500 to 5,000 TEU Intraregional Shanghai–Manila 30 Africa Rotterdam–Genoa 30 Latin America Algeciras–Tema 10 1.00 Oceania Charleston–Santos 10 Transatlantic Xiamen–Brisbane 10 Felixstowe–New York 10 5,000 to 8,500 TEU Africa Hong Kong port–Tema 20 Latin America Charleston–Santos 20 Oceania Xiamen–Brisbane 20 1.54 Transatlantic Felixstowe–New York 20 Asia–Middle East Shanghai–Dubai 20 8,500 to 13,500 TEU Transpacific Busan–Charleston (via Panama) 25 Asia–Middle East Hong Kong port–Los Angeles 25 1.97 Asia–Mediterranean Shanghai–Dubai 25 Singapore–Piraeus 25 Greater than 13,500 TEU Asia–Mediterranean Singapore–Piraeus 40 Asia–North Europe Singapore–Rotterdam 40 2.57 Transpacific Hong Kong port–Los Angeles 20 Source: Original calculations for this publication, based on CPPI 2020 data. The Container Port Performance Index 2020 66 89. The CPPI focuses primarily on micro-delays and assumes these would be recovered on long-haul ocean legs, rather than between coastal ports, which would be more costly. Through simulation, if the index values are tweaked up or down by up to 10 percent, the overall ranking is unaffected. If values are adjusted so that larger ship size groups have lower indices than smaller ones, the overall ranking results can change radically. To achieve a final CPPI score and ranking in the administra- tive approach, accumulated results within each ship size group are multiplied by the index values per ship size group and then summed. The ranking is then based in descending order on final summed totals across all ship size groups. The resulting index for main and secondary ports using the administrative approach is presented in chapter 4 as well as appendix A (table A.2). Constructing the CPPI: The statistical approach 90. An alternative approach to constructing an index of container port performance is to use a statistical technique such as factor analysis. The advantage of factor analysis (FA) is its ability to examine a large dataset and ascertain the impact of a series of measured variables on an unseen latent variable (for example, in this case of efficiency), which cannot be measured directly with a single variable. Instead, the impact is seen through relationships with a series of visible and measurable variables, each of which contains information about the “efficiency” of the port. The latent variable, efficiency, is a function of each of the measured variables and an error term for each. FA then determines the relative weight attached to each of the measured variables, vis-à-vis the efficiency of the port, together with some uncertainty, which is cap- tured by the error terms. 91. At the level of the individual port, the measured variables include the average time expended for the different stages, under various cate- gories, such as different call size bands and berth/port-to-berth. The actual values of these variables are determined by a small number of unobserved factors, such as the availability and quality of the infrastruc- ture, the layout of the port, the expertise of the employees, the avail- able depth in the channel and at the berth, and so on. The challenge lies in the inability to observe these latent factors and how they contribute to the impact on the measured variable, and ultimately on the latent variable. A simple example may be illustrative: Imagine three ports, each with four different types of time cost, as shown in table 3.10. A Comparable Assessment of Container Port Performance 67 Table 3.10. Simple Illustration of Latent Factors Port Cost 1 Cost 2 Cost 3 Cost 4 A 1 2 3 4 B 2 4 6 8 C 3 6 9 12 Source: Original calculations for this publication, based on CPPI 2020 data. 92. As observed in table 3.10, costs 2, 3, and 4 are simply multiples of cost 1. Although each port has four variables, one variable alone is enough to rank the efficiency of these three ports (A > B > C). Of course, this example represents an extreme case, but the idea can be generalized if these variables are correlated to a less extreme extent. In that case, the factors are computed as a linear combination of costs 1 through 4. How- ever, if costs 1 through 4 are completely independent of each other, this method makes no sense. Fortunately, this is not the case for the CPPI dataset. Thus, for each port, we can compute its score on all factors and then combine those scores together to reach a final efficiency score. 93. Note that in the statistical approach using FA, the scores are not calculated for each call size range. On the contrary, the whole data- set, including the smaller ports, is used simultaneously to obtain latent factors. This sharply contrasts with the administrative approach. The statistical approach factors in all the correlations among hours for various call size bands which is—purely from a statistical perspective— more efficient. However, one downside of this approach is that FA does not consider some observations might be more reliable than others as they are based on more calls. This implies the results, and potentially the ranking, for some of the ports could be distorted in the presence of large outliers. 94. Neither methodology is better than the other; rather, the two dif- ferent approaches complement each other. This complementarity encouraged the CPPI team, while deciding on the direction to take in the inaugural edition of the index, to use both approaches—a decision that would help ensure the resulting rankings of container port perfor- mance reflect as closely as possible actual port performance, while also being statistically robust. Chapter 4 presents the CPPI 2020 from both approaches, explaining divergences, where possible. The Container Port Performance Index 2020 68 1. A “hub port” is defined as a port called at by deep-sea mainline container ships and serves as a NOTES transhipment point for smaller outlying, or feeder, ports within its geographical region. Typically, at least 35 percent of its total throughput would be hub and spoke transhipment container activ- ity. In larger hub ports anywhere from 60 to 80 percent of throughput will be transhipment. 2. The list of ports omitted from the CPPI 2020 due to insufficient calls is provided in table A.4 at the end of appendix A. 3. Definitions employed in figure 3.4: • Other berth hours = Activities between all-fast and first lift (“start”) plus the time taken to depart from the berth (all lines released) after the last container lift (“finish”). • Other port hours (“arrival”) = The combination of idle/waiting time at anchorage plus the time required to steam-in from the port limits and until all fast alongside the berth. • Operating hours = The time required for container operations between the first and last con- tainer lifts. 4. Crane intensity is calculated by total accumulated gross crane hours divided by operating hours. 5. The numbers in the green highlighted cells have been imputed by multiplying the base cells by the factor multiplier determined by the overall group average. 6. The precise approach to producing a robust dataset is detailed in appendix B. 7. The actual equation is: (Group Average Port Hours / Example Port Hours) × Call Size Group Weight. A Comparable Assessment of Container Port Performance 69 4.  The Container Port Performance Index 2020 4.1. Introduction 95. The CPPI 2020 has been developed based on total port time in the manner explained in the previous sections. The rationale of using two methodologies, as explained earlier, is to ensure that rankings reflect as closely as possible actual port performance, while also being statis- tically robust. This first inaugural edition utilizes data through June 30, 2020, and while the ranking reflects the performance over the previous twelve months, a qualifier is raised to reflect the impact of the COVID- 19 pandemic on port performance in the first six months of the year. The next edition of the CPPI (CPPI 2021) will offer some comparative analysis of the year-on-year changes in the ranking. The Container Port Performance Index 2020 70 96. This chapter discusses the resulting rankings of container port perfor- mance, based on the two different approaches. Section 4.2 discusses the two approaches, along with the rankings for the best perform- ing container ports. Appendix A includes the full ranking of all ports by each approach separately and presented alphabetically. Table A.1 shares the CPPI 2020 data based on the administrative approach, while table A.2 presents the data based on the statistical approach. To con- clude the report, section 4.3 presents some areas of further work and planned developments for future iterations of the CPPI. Understanding and eliminating divergences between the two methodologies will be part of that agenda. 4.2.  The CPPI 2020 97. Table 4.1 presents the rankings of container port performance in the CPPI 2020, which resulted from using both the statistical and administrative approaches. The ranking and score in the left-hand columns includes results from the statistical approach; the ranking and score in the right-hand columns shows results from the adminis- trative approach. The index points used to construct the ranking in the administrative approach reflect the approach as outlined in the pre- vious chapter, which is an aggregate of the performance of the port, weighted relative to the average, across call and vessel size (see table A.1 for the detailed ranking produced by the administrative approach). Accordingly, the scores can be negative, where a port compares poorly to the average in one call size and vessel size category, particularly if they do not have an offsetting positive score or scores in other cells. 98. The use of factor analysis (FA) also results in a statistic (a total score), that is the sum of a weighted average of indices for each of the same five vessel sizes. The indices for each vessel size are estimated based on the time expired in the port, and a number of unknown factors, or latent variables (see appendix B for a more detailed explanation of the approach), which have an impact on performance, but cannot be seen. The resulting total scores are standardized, with a “negative” score indicating a better than average performance (table A.2 presents the complete rankings generated from the statistical approach). Overall, A Comparable Assessment of Container Port Performance 71 a broad consistency emerges between the rankings that result from the two approaches, with some exceptions. Just under 18 percent of all ports (61 ports) are ranked within three places or less from themselves in the dual rankings provided in table 4.1. Approximately 40 percent (137 ports) are ranked with ten places or less of themselves in the respective ranking, while 80 percent (282 ports) fall within 10 percent of their respective rankings in the two indices. 99. The two top ranked container ports in the CPPI 2020 are Yokohama por (Japan) in first place, followed by King Abdullah Port (Saudi Ara- bia) in second place. These two ports occupy the same positions in the rankings generated by both approaches. East Asia ports dominate the top 50 ranked ports, with several ports in the Middle East and North Africa region, such as King Abdullah port in Saudi Arabia, Salalah in Oman (ranked 6th and 9th in the statistical (or FA) and administrative approaches respectively), Khalifa port in Abu Dhabi (ranked 26th and 22nd respectively), and Tanger Med (ranked at 27th and 15th respec- tively) as the notable exceptions. Algeciras is the highest ranked port in Europe (ranked 10th and 32nd respectively), followed by Aarhus (ranked 44th and 43rd respectively). Colombo is the top-ranked port in South Asia (ranked 17th and 33rd respectively). Lazaro Cardenas the highest ranked port in Latin America (ranked 25th and 23rd respec- tively), with Halifax being the highest ranked port in North America (ranked 39th and 25th respectively). No ports in Sub-Saharan Africa appear in the global top 50 container ports; Djibouti port ranks the highest (61st and 93rd respectively). Table 4.1. The CPPI 2020: Global Ranking of Container Ports Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points YOKOHAMA 1 −5.995 YOKOHAMA 1 130 KING ABDULLAH PORT 2 −5.684 KING ABDULLAH PORT 2 114 CHIWAN 3 −5.202 QINGDAO 3 102 GUANGZHOU 4 −5.162 KAOHSIUNG 4 99 KAOHSIUNG 5 −4.669 SHEKOU 5 94 SALALAH 6 −4.531 GUANGZHOU 6 92 The Container Port Performance Index 2020 72 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points HONG KONG, HONG KONG SAR, CHINA 7 −4.276 HONG KONG, HONG KONG SAR, CHINA 7 89 QINGDAO 8 −3.860 ZHOUSHAN 8 88 SHEKOU 9 −3.726 SALALAH 9 87 ALGECIRAS 10 −3.597 YANGSHAN 10 87 BEIRUT 11 −3.378 TANJUNG PELEPAS 11 86 SHIMIZU 12 −3.361 SINGAPORE 12 83 TANJUNG PELEPAS 13 −3.342 NINGBO 13 83 PORT KLANG 14 −3.334 PORT KLANG 14 78 SINGAPORE 15 −3.279 TANGER MEDITERRANEAN 15 76 NAGOYA 16 −3.251 TAIPEI, TAIWAN, CHINA 16 75 COLOMBO 17 −3.209 YANTIAN 17 73 SINES 18 −3.183 CAI MEP 18 68 KOBE 19 −3.127 DALIAN 19 66 ZHOUSHAN 20 −2.963 TIANJIN 20 64 JUBAIL 21 −2.898 FUZHOU 21 61 YEOSU 22 −2.831 KHALIFA PORT 22 60 FUZHOU 23 −2.829 LAZARO CARDENAS 23 60 NINGBO 24 −2.805 SHIMIZU 24 59 LAZARO CARDENAS 25 −2.798 HALIFAX 25 59 KHALIFA PORT 26 −2.795 XIAMEN 26 58 TANGER MEDITERRANEAN 27 −2.769 CHIWAN 27 58 YANGSHAN 28 −2.733 SINES 28 56 YANTIAN 29 −2.724 MAWAN 29 56 TAIPEI, TAIWAN, CHINA 30 −2.681 AGUADULCE (COLOMBIA) 30 56 DA CHAN BAY TERMINAL ONE 31 −2.588 LIANYUNGANG 31 54 MAWAN 32 −2.557 ALGECIRAS 32 53 DALIAN 33 −2.506 COLOMBO 33 53 INCHEON 34 −2.422 CARTAGENA (COLOMBIA) 34 52 TOKYO 35 −2.418 DA CHAN BAY TERMINAL ONE 35 52 HAMAD PORT 36 −2.411 BUSAN 36 51 A Comparable Assessment of Container Port Performance 73 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points LIANYUNGANG 37 −2.375 INCHEON 37 51 PIPAVAV 38 −2.371 HAMAD PORT 38 51 HALIFAX 39 −2.365 PIPAVAV 39 48 CAUCEDO 40 −2.355 YEOSU 40 48 BREMERHAVEN 41 −2.265 AQABA 41 47 CARTAGENA (COLOMBIA) 42 −2.185 JEDDAH 42 46 SALVADOR 43 −2.051 AARHUS 43 43 AARHUS 44 −2.036 MUNDRA 44 43 AGUADULCE (COLOMBIA) 45 −2.035 MAGDALLA 45 42 CAI LAN 46 −1.991 BARCELONA 46 42 HAIPHONG 47 −1.953 RIO GRANDE (BRAZIL) 47 42 MAGDALLA 48 −1.943 NAGOYA 48 41 CAI MEP 49 −1.932 SHANGHAI 49 41 MUNDRA 50 −1.902 KOBE 50 39 GEMLIK 51 −1.892 LAEM CHABANG 51 39 BUSAN 52 −1.887 WILHELMSHAVEN 52 39 JEDDAH 53 −1.862 CHARLESTON 53 38 DILISKELESI 54 −1.842 TOKYO 54 38 LAEM CHABANG 55 −1.807 SALVADOR 55 35 JAWAHARLAL NEHRU PORT 56 −1.786 CALLAO 56 34 AMBARLI 57 −1.783 JUBAIL 57 33 PORT SAID 58 −1.652 YARIMCA 58 33 PECEM 59 −1.647 JEBEL ALI 59 33 AQABA 60 −1.594 HAIPHONG 60 33 DJIBOUTI 61 −1.590 AMBARLI 61 32 XIAMEN 62 −1.541 BEIRUT 62 32 SHANGHAI 63 −1.532 JAWAHARLAL NEHRU PORT 63 31 TANJUNG PRIOK 64 −1.521 KEELUNG 64 31 KEELUNG 65 −1.509 ANTWERP 65 31 TRIPOLI (LEBANON) 66 −1.497 TANJUNG PRIOK 66 31 The Container Port Performance Index 2020 74 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points OSAKA 67 −1.440 WILMINGTON (NORTH CAROLINA, USA) 67 30 YARIMCA 68 −1.393 CAUCEDO 68 30 ITAPOA 69 −1.376 SOHAR 69 29 SANTOS 70 −1.376 PORT SAID 70 29 SOHAR 71 −1.375 BUENAVENTURA 71 29 BUENAVENTURA 72 −1.353 SANTOS 72 28 SEPETIBA 73 −1.338 BOSTON (USA) 73 27 RIO GRANDE (BRAZIL) 74 −1.332 CAI LAN 74 27 KARACHI 75 −1.292 DILISKELESI 75 27 BARCELONA 76 −1.224 TRIPOLI (LEBANON) 76 27 POSORJA 77 −1.192 OSAKA 77 27 OSLO 78 −1.192 BALBOA 78 27 QUY NHON 79 −1.163 GEMLIK 79 26 CAT LAI 80 −1.149 KARACHI 80 26 SUAPE 81 −1.129 YOKKAICHI 81 26 LONDON 82 −1.117 COLON 82 25 PHILADELPHIA 83 −1.080 MERSIN 83 25 DANANG 84 −1.053 SEPETIBA 84 24 PORT OF VIRGINIA 85 −1.044 BREMERHAVEN 85 24 ANTWERP 86 −1.011 PECEM 86 23 ZEEBRUGGE 87 −0.988 PHILADELPHIA 87 23 SANTA CRUZ DE TENERIFE 88 −0.970 ITAPOA 88 23 NEW YORK & NEW JERSEY 89 −0.969 PIRAEUS 89 23 GDYNIA 90 −0.942 ROTTERDAM 90 22 SHANTOU 91 −0.935 OSLO 91 20 NAHA 92 −0.883 DAMMAM 92 20 PIRAEUS 93 −0.859 DJIBOUTI 93 19 PUERTO LIMON 94 −0.858 PAITA 94 19 CHARLESTON 95 −0.820 SANTA CRUZ DE TENERIFE 95 19 PORT AKDENIZ 96 −0.820 SAVANNAH 96 18 A Comparable Assessment of Container Port Performance 75 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points TAICHUNG 97 −0.814 VERACRUZ 97 18 ULSAN 98 −0.811 ALTAMIRA 98 17 HAKATA 99 −0.758 JACKSONVILLE 99 17 COLON 100 −0.752 TAURANGA 100 17 MOBILE 101 −0.745 MARSAXLOKK 101 16 DAMMAM 102 −0.737 VALPARAISO 102 16 PUERTO BARRIOS 103 −0.731 GDYNIA 103 16 NOUMEA 104 −0.721 QUY NHON 104 16 PAITA 105 −0.705 ITAJAI 105 16 MARSAXLOKK 106 −0.698 PUERTO LIMON 106 16 YOKKAICHI 107 −0.692 PARANAGUA 107 16 MALAGA 108 −0.690 JOHOR 108 15 OMAEZAKI 109 −0.680 CAT LAI 109 15 JOHOR 110 −0.669 PORT OF VIRGINIA 110 15 MOJI 111 −0.663 NAHA 111 15 BATANGAS 112 −0.663 FORT−DE−FRANCE 112 15 BOSTON (USA) 113 −0.631 POINTE−A−PITRE 113 14 BALBOA 114 −0.625 MIAMI 114 14 SIAM SEAPORT 115 −0.613 DANANG 115 13 ROTTERDAM 116 −0.611 HAKATA 116 12 BURGAS 117 −0.611 SHANTOU 117 12 DUNKIRK 118 −0.611 AUCKLAND 118 12 WILHELMSHAVEN 119 −0.598 BALTIMORE (USA) 119 12 SANTA MARTA 120 −0.589 DAKAR 120 12 TARRAGONA 121 −0.588 OMAEZAKI 121 11 CAGAYAN DE ORO 122 −0.582 PUERTO BARRIOS 122 11 AUCKLAND 123 −0.562 MOJI 123 11 NORRKOPING 124 −0.553 NOUMEA 124 11 FORT−DE−FRANCE 125 −0.543 ENSENADA 125 11 GUSTAVIA 126 −0.528 POSORJA 126 10 The Container Port Performance Index 2020 76 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points QINZHOU 127 −0.523 NEW YORK & NEW JERSEY 127 10 FREDERICIA 128 −0.493 CAGAYAN DE ORO 128 10 COPENHAGEN 129 −0.471 MALAGA 129 10 TUTICORIN 130 −0.469 SUAPE 130 10 PUERTO CORTES 131 −0.468 GDANSK 131 10 PORT BRONKA 132 −0.454 MOBILE 132 10 TANJUNG EMAS 133 −0.433 BURGAS 133 9 CHIBA 134 −0.418 TARRAGONA 134 9 SHIBUSHI 135 −0.382 NORRKOPING 135 9 ENSENADA 136 −0.378 SIAM SEAPORT 136 9 SAIGON 137 −0.375 SANTA MARTA 137 9 BALTIMORE (USA) 138 −0.368 FREDERICIA 138 9 PENANG 139 −0.367 TUTICORIN 139 8 KHALIFA BIN SALMAN 140 −0.356 CEBU 140 8 PUERTO BOLIVAR (ECUADOR) 141 −0.347 WELLINGTON 141 8 GIOIA TAURO 142 −0.344 COPENHAGEN 142 8 DAVAO 143 −0.341 TANJUNG EMAS 143 8 CASTELLON 144 −0.339 BATANGAS 144 8 ALTAMIRA 145 −0.324 MANZANILLO (MEXICO) 145 8 RIO HAINA 146 −0.322 TAICHUNG 146 8 POINTE−A−PITRE 147 −0.319 PORT EVERGLADES 147 8 SAN JUAN 148 −0.303 QINZHOU 148 8 SHUAIBA 149 −0.301 PUERTO CORTES 149 7 NAPLES 150 −0.280 CHIBA 150 7 CEBU 151 −0.275 SAIGON 151 7 FREEPORT (BAHAMAS) 152 −0.271 PORT BRONKA 152 7 TANJUNG PERAK 153 −0.269 HAIFA 153 6 PUERTO QUETZAL 154 −0.269 PORT AKDENIZ 154 6 WILMINGTON (NORTH CAROLINA, USA) 155 −0.237 PUERTO BOLIVAR (ECUADOR) 155 6 A Comparable Assessment of Container Port Performance 77 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points VIGO 156 −0.233 DAVAO 156 6 PAPEETE 157 −0.227 CASTELLON 157 6 VERACRUZ 158 −0.219 CORONEL 158 6 SAN ANTONIO 159 −0.218 VIGO 159 5 SHUWAIKH 160 −0.216 RIO HAINA 160 5 JACKSONVILLE 161 −0.215 GUSTAVIA 161 5 CALDERA (COSTA RICA) 162 −0.206 KHALIFA BIN SALMAN 162 5 BELL BAY 163 −0.205 BELL BAY 163 5 HELSINGBORG 164 −0.196 SAN JUAN 164 5 POINT LISAS PORTS 165 −0.182 CALDERA (COSTA RICA) 165 4 BORUSAN 166 −0.163 SHIBUSHI 166 4 KALININGRAD 167 −0.159 PAPEETE 167 4 CARTAGENA (SPAIN) 168 −0.153 SALERNO 168 3 BARRANQUILLA 169 −0.137 BARRANQUILLA 169 3 CRISTOBAL 170 −0.135 YUZHNY 170 3 LATAKIA 171 −0.135 HELSINGBORG 171 3 PALERMO 172 −0.125 TANJUNG PERAK 172 3 NASSAU 173 −0.109 SAN ANTONIO 173 3 SALERNO 174 −0.080 PUERTO QUETZAL 174 2 CIVITAVECCHIA 175 −0.072 ULSAN 175 2 WELLINGTON 176 −0.049 RAVENNA 176 2 DAKAR 177 −0.031 POINT LISAS PORTS 177 2 HERAKLION 178 −0.030 LIMASSOL 178 1 RAVENNA 179 −0.025 CARTAGENA (SPAIN) 179 1 CATANIA 180 −0.021 LONDON 180 1 LARVIK 181 −0.015 CIVITAVECCHIA 181 1 PORT FREEPORT 182 −0.008 PALERMO 182 1 VILA DO CONDE 183 0.000 ANCONA 183 0 VITORIA 183 0.000 LARVIK 184 0 The Container Port Performance Index 2020 78 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points OITA 185 0.004 HERAKLION 185 0 KRISTIANSAND 186 0.008 LYTTELTON 186 0 GIJON 187 0.013 LATAKIA 187 0 LYTTELTON 188 0.034 OITA 188 0 ODESSA 189 0.046 NASSAU 189 0 DURRES 190 0.057 CATANIA 190 0 MARIEL 191 0.057 HAMBURG 191 0 LIMASSOL 192 0.063 PHILIPSBURG 192 0 ANCONA 193 0.067 PORT FREEPORT 193 0 PHILIPSBURG 194 0.090 KRISTIANSAND 194 −1 MUUGA—PORT OF TALLINN 195 0.092 BORUSAN 195 −1 VARNA 196 0.094 GIJON 196 −1 BELAWAN 197 0.104 MARIEL 197 −1 YUZHNY 198 0.105 MUUGA—PORT OF TALLINN 198 −1 RIO DE JANEIRO 199 0.126 SHUWAIKH 199 −1 MIAMI 200 0.149 TAMPA 200 −1 BARI 201 0.167 SANTO TOMAS DE CASTILLA 201 −2 VLADIVOSTOK 202 0.182 BELAWAN 202 −2 SANTO TOMAS DE CASTILLA 203 0.185 BARI 203 −2 VOSTOCHNY 204 0.192 TOMAKOMAI 204 −2 TOMAKOMAI 205 0.194 MATADI 205 −2 LEIXOES 206 0.201 NELSON 206 −2 ARICA 207 0.211 NAPLES 207 −2 PORT EVERGLADES 208 0.213 RAUMA 208 −3 MATADI 209 0.238 POTI 209 −3 NELSON 210 0.240 LEIXOES 210 −3 SAINT JOHN 211 0.241 PORT MORESBY 211 −4 TRIESTE 212 0.241 GIOIA TAURO 212 −4 GAVLE 213 0.242 VARNA 213 −4 A Comparable Assessment of Container Port Performance 79 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points MERSIN 214 0.275 GAVLE 214 −4 POTI 215 0.278 NEW MANGALORE 215 −4 TEESPORT 216 0.303 FREETOWN 216 −5 PORT MORESBY 217 0.307 ABIDJAN 217 −5 TIANJIN 218 0.310 VITORIA 218 −5 TAURANGA 219 0.333 TOAMASINA 219 −5 FREETOWN 220 0.350 KALININGRAD 220 −5 SAN VICENTE 221 0.361 ALEXANDRIA (EGYPT) 221 −5 VALPARAISO 222 0.365 GRANGEMOUTH 222 −6 BERBERA 223 0.368 RIGA 223 −6 BILBAO 224 0.384 FREEPORT (BAHAMAS) 224 −6 HAIFA 225 0.384 ZEEBRUGGE 225 −6 TAMPA 226 0.411 SAN VICENTE 226 −7 TOAMASINA 227 0.433 PENANG 227 −7 ABIDJAN 228 0.439 SHUAIBA 228 −7 RAUMA 229 0.439 PANJANG 229 −7 CHORNOMORSK 230 0.449 KOPER 230 −7 CALLAO 231 0.450 DURRES 231 −7 NEW MANGALORE 232 0.451 APRA HARBOR 232 −7 UMM QASR 233 0.452 COTONOU 233 −8 ALEXANDRIA (EGYPT) 234 0.482 BRISBANE 234 −8 RIGA 235 0.482 TAKORADI 235 −8 SEATTLE 236 0.487 PUERTO PROGRESO 236 −8 VANCOUVER (CANADA) 237 0.499 NEW ORLEANS 237 −9 CONAKRY 238 0.505 TEESPORT 238 −9 EL DEKHEILA 239 0.509 BLUFF 239 −9 NAPIER 240 0.513 VLADIVOSTOK 240 −9 LA SPEZIA 241 0.548 VILA DO CONDE 241 −9 KOTKA 242 0.558 HOUSTON 242 −9 The Container Port Performance Index 2020 80 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points GRANGEMOUTH 243 0.560 BILBAO 243 −10 COTONOU 244 0.562 KOTKA 244 −10 OWENDO 245 0.563 BERBERA 245 −10 BRISBANE 246 0.569 NANTES−ST NAZAIRE 246 −10 PANJANG 247 0.573 ARICA 247 −11 APRA HARBOR 248 0.610 PORT OF SPAIN 248 −11 ACAJUTLA 249 0.640 ACAJUTLA 249 −11 DAMIETTA 250 0.650 CONAKRY 250 −11 LIVORNO 251 0.658 SAINT JOHN 251 −12 PARANAGUA 252 0.659 DUBLIN 252 −12 CONSTANTZA 253 0.660 STOCKHOLM 253 −13 TAKORADI 254 0.683 MALABO 254 −14 NANTES−ST NAZAIRE 255 0.693 VOSTOCHNY 255 −14 LIRQUEN 256 0.708 EL DEKHEILA 256 −14 BLUFF 257 0.731 CRISTOBAL 257 −14 LE HAVRE 258 0.746 ST PETERSBURG 258 −14 PUERTO PROGRESO 259 0.750 BANGKOK 259 −15 PORT OF SPAIN 260 0.775 MOGADISCIO 260 −15 TEMA 261 0.782 NAPIER 261 −15 DUBLIN 262 0.839 LIRQUEN 262 −15 MALABO 263 0.859 SEATTLE 263 −16 HUENEME 264 0.862 TRIESTE 264 −16 NEW ORLEANS 265 0.865 ODESSA 265 −17 HOUSTON 266 0.878 KLAIPEDA 266 −17 STOCKHOLM 267 0.915 AGADIR 267 −17 BEIRA 268 0.919 CASABLANCA 268 −17 MANAUS 269 0.967 IZMIR 269 −18 ONNE 270 0.994 MANAUS 270 −18 BANGKOK 271 1.024 OWENDO 271 −18 A Comparable Assessment of Container Port Performance 81 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points ISKENDERUN 272 1.024 ISKENDERUN 272 −18 MONTEVIDEO 273 1.033 HUENEME 273 −19 VENICE 274 1.042 CHORNOMORSK 274 −19 MANZANILLO (MEXICO) 275 1.070 BATUMI 275 −19 AGADIR 276 1.122 MANILA 276 −19 IZMIR 277 1.136 BRISTOL 277 −19 GENERAL SANTOS 278 1.148 LAE 278 −19 SAVANNAH 279 1.158 RIJEKA 279 −20 HAMBURG 280 1.176 SAMSUN 280 −20 MOGADISCIO 281 1.194 BEIRA 281 −21 CORONEL 282 1.203 GENERAL SANTOS 282 −21 VALENCIA 283 1.211 MONTREAL 283 −22 THESSALONIKI 284 1.229 THESSALONIKI 284 −22 MONTREAL 285 1.231 CONSTANTZA 285 −22 KINGSTON (JAMAICA) 286 1.283 KINGSTON (JAMAICA) 286 −23 LAE 287 1.314 KRIBI DEEP SEA PORT 287 −24 LOME 288 1.332 OTAGO HARBOUR 288 −24 MEJILLONES 289 1.381 LA SPEZIA 289 −24 SOUTHAMPTON 290 1.404 LIVORNO 290 −24 CASABLANCA 291 1.442 ONNE 291 −25 PORT VICTORIA 292 1.457 NOUAKCHOTT 292 −25 SOKHNA 293 1.459 SAN PEDRO (COTE D'IVOIRE) 293 −27 BRISTOL 294 1.462 UMM QASR 294 −27 NOUAKCHOTT 295 1.475 VENICE 295 −27 MUHAMMAD BIN QASIM 296 1.499 LA GUAIRA 296 −28 SAMSUN 297 1.502 DAMIETTA 297 −28 DOUALA 298 1.510 CHATTOGRAM 298 −28 MAPUTO 299 1.533 PORT VICTORIA 299 −28 NOVOROSSIYSK 300 1.626 LE HAVRE 300 −30 The Container Port Performance Index 2020 82 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points KLAIPEDA 301 1.635 NOVOROSSIYSK 301 −30 MELBOURNE 302 1.676 DOUALA 302 −31 KRIBI DEEP SEA PORT 303 1.701 VANCOUVER (CANADA) 303 −31 ST PETERSBURG 304 1.719 TIMARU 304 −32 LA GUAIRA 305 1.737 DUNKIRK 305 −33 CHATTOGRAM 306 1.809 MAPUTO 306 −33 WALVIS BAY 307 1.819 PUERTO CABELLO 307 −34 ITAJAI 308 1.828 VALENCIA 308 −34 PUERTO CABELLO 309 1.829 RIO DE JANEIRO 309 −35 SAVONA−VADO 310 1.897 BUENOS AIRES 310 −37 GDANSK 311 1.944 MEJILLONES 311 −38 RIJEKA 312 1.974 TEMA 312 −40 FELIXSTOWE 313 2.006 MELBOURNE 313 −40 OTAGO HARBOUR 314 2.023 SAVONA−VADO 314 −40 ALGIERS 315 2.076 ASHDOD 315 −42 BATUMI 316 2.175 MUHAMMAD BIN QASIM 316 −42 TIMARU 317 2.225 SOUTHAMPTON 317 −45 KOPER 318 2.237 ALGIERS 318 −49 SAN PEDRO (COTE D'IVOIRE) 319 2.267 FREMANTLE 319 −49 BUENOS AIRES 320 2.391 IQUIQUE 320 −54 GENOA 321 2.420 SOKHNA 321 −55 MANILA 322 2.445 FELIXSTOWE 322 −55 JEBEL ALI 323 2.482 PRINCE RUPERT 323 −56 DAR ES SALAAM 324 2.561 DAR ES SALAAM 324 −58 DUTCH HARBOR 325 2.591 DUTCH HARBOR 325 −60 FREMANTLE 326 2.716 NEMRUT BAY 326 −61 ASHDOD 327 2.797 PORT BOTANY 327 −63 LOS ANGELES 328 2.899 MONTEVIDEO 328 −65 NEMRUT BAY 329 2.970 TACOMA 329 −66 PRINCE RUPERT 330 2.979 BEJAIA 330 −72 A Comparable Assessment of Container Port Performance 83 Statistical approach Administrative approach Port name Rank Total score Port name Rank Index points MOMBASA 331 3.140 GENOA 331 −74 OAKLAND 332 3.163 LOME 332 −77 LONG BEACH 333 3.175 ADELAIDE 333 −78 PORT REUNION 334 3.302 OAKLAND 334 −79 TACOMA 335 3.628 MOMBASA 335 −79 GUAYAQUIL 336 3.647 WALVIS BAY 336 −80 PORT BOTANY 337 3.907 LOS ANGELES 337 −82 BEJAIA 338 4.054 GUAYAQUIL 338 −84 ADELAIDE 339 4.546 GOTHENBURG 339 −87 LAGOS (NIGERIA) 340 4.646 PORT REUNION 340 −89 GOTHENBURG 341 4.653 LONG BEACH 341 −96 IQUIQUE 342 4.766 LAGOS (NIGERIA) 342 −114 TIN CAN ISLAND 343 4.789 LUANDA 343 −115 PORT LOUIS 344 5.501 TIN CAN ISLAND 344 −118 MARSEILLE 345 5.696 POINTE−NOIRE 345 −128 POINTE−NOIRE 346 5.832 PORT LOUIS 346 −175 CAPE TOWN 347 6.528 CAPE TOWN 347 −177 PORT ELIZABETH 348 7.659 PORT ELIZABETH 348 −183 DURBAN 349 8.082 NGQURA 349 −190 LUANDA 350 8.383 MARSEILLE 350 −238 NGQURA 351 8.401 DURBAN 351 −255 100. Conversely, the rankings reflect some significant discrepancies in a limited number of ports. Three of the most notable discrepancies are Jebel Ali (323rd and 59th in the FA and administrative approach respectively), Itajai (308th and 105th respectively), and Tianjin, one of the highest ranked ports affected (218th and 20th respectively). Rather than explaining every discrepancy, the approach taken in the CPPI 2020 attempts to make the methodology and assumptions explicit and let the data speak. In future editions, further refinement of both approaches will hopefully reduce discrepancies, and further enhance their respective complementarity, and the rigor of the CPPI. The Container Port Performance Index 2020 84 4.3.  Conclusions and Next Steps 101. The rationale behind the CPPI focuses on using the available empir- ical data to create an objective measure to then compare container port performance across ports, and eventually over time. The inten- tion is to, in a comparable manner, identify gaps and opportunities for improvement that will ultimately benefit all stakeholders, from shipping lines to national governments to consumers. The CPPI is intended to serve as a key reference point for stakeholders in the global economy, including national governments, port authorities and operators, development agencies, supranational organizations, vari- ous maritime interests, and other public and private stakeholders in trade, logistics, and supply chain services. 102. Looking to the future, the intention is that the CPPI will evolve and be enhanced in subsequent editions, reflecting refinement, stake- holder feedback, and improvement in data scope and quality. The WBG/IHS Markit team will continue to refine the methodologies and scope, increasing the number of ports as well as the data where pos- sible. The next iteration (CPPI 2021) will be comparable, facilitating the introduction of trends in container port performance, both overall and potentially by disaggregation by ship or call size. The CPPI 2021 will also seek to investigate and explain divergences between the two approaches, while also gaining a further understanding of key deter- minants or influences on container port performance. The overall objective remains the identification of opportunities for improve- ment, which will ultimately benefit all public and private stakeholders in the sector. A Comparable Assessment of Container Port Performance 85 Appendix A: The CPPI 2020 Table A.1. The CPPI 2020: Administrative Approach Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points AARHUS 43 43.45 17 7.1 28 10.0 — — 18 11.7 ABIDJAN 217 (4.67) — 222 (1.8) 149 (1.9) — — ACAJUTLA 249 (11.08) — 287 (11.1) — — — ADELAIDE 333 (78.35) — 242 (3.6) 205 (22.4) 150 (20.4) — AGADIR 267 (17.01) 177 (4.4) 302 (15.0) — — — AGUADULCE 30 55.65 — 80 7.3 — 13 12.3 25 9.4 (COLOMBIA) ALEXANDRIA 221 (5.49) 196 (7.2) 226 (2.2) — — — (EGYPT) ALGECIRAS 32 52.90 83 2.7 118 5.3 75 6.6 44 8.3 30 7.8 ALGIERS 318 (48.92) 207 (9.7) 326 (44.5) — — — ALTAMIRA 98 16.77 43 4.6 73 7.7 95 4.5 — — The Container Port Performance Index 2020 86 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points AMBARLI 61 32.35 86 2.4 161 3.0 77 6.2 45 8.2 61 1.0 ANCONA 183 0.48 70 3.4 214 (1.1) — — — ANTWERP 65 30.76 104 1.4 157 3.2 101 3.7 72 4.5 43 4.8 APRA HARBOR 232 (7.46) 178 (4.5) 254 (5.4) — — — AQABA 41 46.92 — 108 6.0 133 0.2 34 9.4 27 8.6 ARICA 247 (10.55) 168 (3.3) 198 (0.3) 163 (5.7) — — ASHDOD 315 (41.91) 174 (4.2) 282 (10.1) 187 (12.7) — 72 (4.0) AUCKLAND 118 11.70 — 103 6.2 104 3.6 — — BALBOA 78 26.58 131 (0.5) 156 3.4 117 2.3 69 4.7 47 4.1 BALTIMORE 119 11.64 — 179 1.4 132 0.6 71 4.7 — (USA) BANGKOK 259 (14.58) 191 (6.1) 289 (11.8) — — — BARCELONA 46 41.83 89 2.3 92 6.6 57 8.5 51 7.2 54 2.7 BARI 203 (1.92) 136 (0.9) 220 (1.5) — — — BARRANQUILLA 169 3.29 72 3.2 174 1.8 — — — BATANGAS 144 7.88 — 66 7.9 — — — BATUMI 275 (18.89) 147 (1.5) 307 (18.2) — — — BEIRA 281 (20.59) — 313 (20.6) — — — BEIRUT 62 31.86 36 5.3 45 8.8 64 8.1 70 4.7 65 (0.4) BEJAIA 330 (71.85) 219 (27.7) 327 (59.1) — — — BELAWAN 202 (1.77) 119 0.2 223 (1.9) — — — BELL BAY 163 4.84 87 2.3 150 3.8 — — — BERBERA 245 (9.98) — 281 (10.0) — — — BILBAO 243 (9.58) 143 (1.3) 275 (9.0) — — — BLUFF 239 (8.62) — 272 (8.6) — — — BORUSAN 195 (0.67) — 208 (0.7) — — — BOSTON (USA) 73 27.47 — 71 7.7 61 8.2 80 3.6 — BREMERHAVEN 85 23.53 109 0.6 146 3.9 102 3.6 89 3.0 52 3.0 A Comparable Assessment of Container Port Performance 87 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points BRISBANE 234 (7.78) 140 (1.1) 178 1.6 156 (4.6) 112 (0.9) — BRISTOL 277 (19.33) 206 (9.6) 301 (14.9) — — — BUENAVENTURA 71 28.58 — 68 7.8 35 9.4 88 3.2 — BUENOS AIRES 310 (36.93) — 207 (0.6) 153 (3.7) 144 (15.5) — BURGAS 133 9.46 81 2.9 56 8.1 — — — BUSAN 36 51.23 38 5.1 70 7.8 59 8.3 101 1.0 21 10.3 CAGAYAN DE 128 9.92 53 4.0 60 8.1 — — — ORO CAI LAN 74 27.43 — 14 11.2 18 10.5 — — CAI MEP 18 67.63 — 5 12.1 22 10.4 48 7.4 24 9.7 CALDERA 165 4.07 59 3.7 167 2.4 — — — (COSTA RICA) CALLAO 56 34.14 106 1.1 202 (0.4) 119 2.0 93 2.1 20 10.4 CAPE TOWN 347 (177.22) 121 (0.0) 305 (17.7) 209 (34.1) 161 (54.3) — CARTAGENA 34 52.26 57 3.8 18 10.8 8 11.4 20 11.2 — (COLOMBIA) CARTAGENA 179 1.39 75 3.0 — — — — (SPAIN) CASABLANCA 268 (17.29) 194 (6.6) 294 (12.8) 141 (0.9) — — CASTELLON 157 5.83 114 0.3 140 4.0 127 1.1 — — CAT LAI 109 15.37 11 8.3 10 11.6 — — — CATANIA 190 (0.20) 129 (0.4) — — — — CAUCEDO 68 29.64 99 1.8 55 8.2 80 5.9 62 5.8 — CEBU 140 8.21 24 6.5 119 5.2 — — — CHARLESTON 53 38.00 — 109 5.8 78 6.1 74 4.4 39 5.5 CHATTOGRAM 298 (27.94) — 320 (27.9) — — — CHIBA 150 7.05 — 85 7.0 — — — CHIWAN 27 57.96 9 8.5 27 10.1 29 9.7 21 11.1 53 2.8 CHORNOMORSK 274 (18.88) 62 3.7 218 (1.4) 186 (12.4) — — The Container Port Performance Index 2020 88 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points CIVITAVECCHIA 181 0.84 97 1.8 — — — — COLOMBO 33 52.51 88 2.3 131 4.6 70 7.3 32 9.6 35 6.5 COLON 82 25.37 80 2.9 91 6.7 129 1.1 91 2.2 45 4.4 CONAKRY 250 (11.26) 151 (1.7) 284 (10.5) — — — CONSTANTZA 285 (22.18) 63 3.7 194 0.5 160 (5.1) 132 (8.4) — COPENHAGEN 142 7.92 71 3.3 99 6.4 — — — CORONEL 158 5.53 — 117 5.3 184 (11.6) 97 1.4 38 5.9 COTONOU 233 (7.73) 60 3.7 259 (6.6) 148 (1.8) — — CRISTOBAL 257 (14.15) — 183 1.3 154 (4.0) 125 (4.7) — DA CHAN BAY 35 52.25 — 30 9.9 13 11.1 8 12.8 — TERMINAL ONE DAKAR 120 11.51 115 0.3 190 0.7 72 7.0 — — DALIAN 19 66.24 — 132 4.6 54 8.6 5 13.1 26 8.8 DAMIETTA 297 (27.91) 149 (1.6) 257 (6.0) 146 (1.7) 76 4.3 85 (10.5) DAMMAM 92 20.03 — 130 4.7 109 3.1 85 3.4 60 1.5 DANANG 115 12.84 2 10.5 63 8.0 — — — DAR ES SALAAM 324 (57.80) — 324 (30.8) 201 (17.5) — — DAVAO 156 5.87 154 (1.7) 258 (6.6) 52 8.6 — — DILISKELESI 75 27.19 14 7.6 35 9.5 39 9.2 — — DJIBOUTI 93 19.43 73 3.2 112 5.6 112 2.8 83 3.4 62 0.5 DOUALA 302 (30.79) 193 (6.6) 319 (27.8) — — — DUBLIN 252 (12.14) 204 (9.5) 269 (7.8) — — — DUNKIRK 305 (32.53) 55 4.0 164 2.5 40 9.1 131 (7.9) 88 (13.7) DURBAN 351 (255.28) 200 (7.8) 312 (19.2) 194 (16.1) 158 (30.6) 98 (57.4) DURRES 231 (7.38) 135 (0.9) 261 (7.0) — — — DUTCH HARBOR 325 (59.57) — 308 (18.3) 206 (26.8) — — EL DEKHEILA 256 (13.98) 96 1.9 200 (0.3) 177 (9.4) — — ENSENADA 125 10.61 — 143 3.9 115 2.6 98 1.3 — FELIXSTOWE 322 (54.78) 148 (1.5) 241 (3.6) 164 (5.7) 129 (7.1) 86 (10.8) A Comparable Assessment of Container Port Performance 89 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points FORT- 112 14.51 157 (1.8) 94 6.5 83 5.7 — — DE-FRANCE FREDERICIA 138 8.58 51 4.1 90 6.7 — — — FREEPORT 224 (6.29) — 114 5.5 144 (1.2) 126 (5.0) — (BAHA-MAS) FREETOWN 216 (4.55) 170 (3.7) 233 (2.9) — — — FREMANTLE 319 (49.22) — 210 (0.9) 171 (7.5) 146 (18.7) — FUZHOU 21 60.89 — 67 7.8 47 8.9 19 11.3 34 6.7 GAVLE 214 (4.02) 126 (0.3) 244 (3.9) — — — GDANSK 131 9.82 44 4.6 171 2.1 — — 58 2.2 GDYNIA 103 16.17 48 4.3 95 6.5 89 5.0 — — GEMLIK 79 26.35 26 5.9 31 9.9 45 8.9 — — GENERAL 282 (21.35) — 314 (21.3) — — — SANTOS GENOA 331 (73.70) 112 0.3 187 0.9 166 (6.3) 134 (8.8) 90 (18.5) GIJON 196 (0.86) 111 0.3 213 (1.0) — — — GIOIA TAURO 212 (3.62) — — 145 (1.7) 100 1.0 69 (1.2) GOTHENBURG 339 (86.64) 125 (0.3) 193 0.5 — — 96 (33.8) GRANGEMOUTH 222 (5.67) 213 (12.3) — — — — GUANGZHOU 6 92.29 46 4.4 6 11.9 12 11.2 9 12.8 10 13.9 GUAYAQUIL 338 (84.32) — 225 (2.2) 183 (11.0) 133 (8.7) 91 (18.7) GUSTAVIA 161 5.10 1 11.1 — — — — HAIFA 153 6.39 127 (0.4) 231 (2.8) 165 (5.9) 110 (0.7) 31 7.7 HAIPHONG 60 32.52 107 1.1 163 2.6 56 8.5 43 8.3 — HAKATA 116 12.45 79 2.9 15 11.1 — — — HALIFAX 25 58.64 76 3.0 75 7.6 79 6.0 46 7.8 22 9.8 HAMAD PORT 38 50.55 — 38 9.3 25 10.1 26 10.2 57 2.2 HAMBURG 191 (0.23) 134 (0.7) 217 (1.4) 131 0.7 99 1.3 67 (0.8) HELSINGBORG 171 3.20 85 2.5 172 2.0 — — — The Container Port Performance Index 2020 90 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points HERAKLION 185 0.09 118 0.2 — — — — HONG KONG 7 88.78 15 7.5 22 10.7 5 12.5 15 11.8 14 12.5 PORT, HONG KONG SAR, CHINA HOUSTON 242 (9.33) — 199 (0.3) 137 (0.5) 122 (4.2) — HUENEME 273 (18.87) — 311 (18.9) — — — INCHEON 37 50.90 19 6.9 51 8.6 38 9.2 10 12.6 — IQUIQUE 320 (54.39) 34 5.4 227 (2.5) — 156 (27.6) — ISKENDERUN 272 (18.49) 28 5.9 93 6.5 68 7.6 149 (20.1) — ITAJAI 105 15.79 — 134 4.5 16 10.7 118 (2.7) — ITAPOA 88 23.05 — 180 1.4 67 7.7 67 5.0 — IZMIR 269 (17.51) 164 (2.5) 240 (3.6) 176 (8.3) — — JACKSONVILLE 99 16.65 — — 85 5.6 78 4.1 — JAWAHARLAL 63 31.34 — 72 7.7 84 5.7 47 7.5 — NEHRU PORT JEBEL ALI 59 32.62 — 230 (2.7) 50 8.8 60 6.1 49 3.8 JEDDAH 42 45.79 176 (4.3) 153 3.6 48 8.8 73 4.5 28 8.5 JOHOR 108 15.43 84 2.7 152 3.6 73 6.9 — — JUBAIL 57 33.35 — — 81 5.9 56 6.5 46 4.4 KALININGRAD 220 (5.46) — 255 (5.5) — — — KAOHSIUNG 4 99.09 8 8.5 11 11.5 27 9.9 17 11.5 3 17.8 KARACHI 80 26.01 — 44 8.9 103 3.6 61 5.9 — KEELUNG 64 30.92 32 5.5 39 9.1 65 8.1 82 3.5 — KHALIFA BIN 162 4.98 — 135 4.4 — 106 0.3 — SALMAN KHALIFA PORT 22 59.99 — 113 5.5 30 9.7 49 7.4 23 9.7 KING 2 114.31 — 2 15.1 6 12.5 4 13.9 2 20.5 ABDULLAH PORT A Comparable Assessment of Container Port Performance 91 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points KINGSTON 286 (22.76) 167 (3.3) 139 4.0 49 8.8 119 (3.1) 87 (12.8) (JAMAICA) KLAIPEDA 266 (16.79) 58 3.7 309 (18.5) — — — KOBE 50 39.40 10 8.3 12 11.5 66 7.8 59 6.2 — KOPER 230 (6.94) 91 2.1 170 2.3 142 (1.0) 90 3.0 75 (5.7) KOTKA 244 (9.87) 163 (2.3) 273 (8.8) — — — KRIBI DEEP SEA 287 (23.64) 217 (20.3) 300 (14.3) — — — PORT KRISTIANSAND 194 (0.64) 145 (1.4) — — — — LA GUAIRA 296 (27.60) 209 (9.8) 316 (23.1) — — — LA SPEZIA 289 (24.02) 146 (1.4) 267 (7.6) 138 (0.5) 115 (1.9) 73 (4.3) LAE 278 (19.34) 210 (11.2) 299 (14.2) — — — LAEM 51 39.04 22 6.5 42 8.9 100 3.9 55 6.6 51 3.1 CHABANG LAGOS 342 (113.72) — 321 (28.1) 212 (55.6) — — (NIGERIA) LARVIK 184 0.11 117 0.2 — — — — LATAKIA 187 0.04 152 (1.7) 189 0.8 — — — LAZARO 23 59.92 — 62 8.0 9 11.4 40 8.7 33 6.7 CARDENAS LE HAVRE 300 (29.67) 139 (1.1) 192 0.6 152 (2.0) 127 (5.7) 76 (6.0) LEIXOES 210 (3.49) 183 (4.8) 215 (1.3) — — — LIANYUNGANG 31 54.38 — 98 6.4 10 11.3 36 9.1 42 4.9 LIMASSOL 178 1.41 169 (3.5) 149 3.8 135 (0.5) — — LIRQUEN 262 (15.33) — 120 5.1 134 (0.0) 135 (10.3) — LIVORNO 290 (24.04) 197 (7.5) 209 (0.9) 188 (12.8) — — LOME 332 (77.24) — 270 (7.8) 211 (45.1) — — LONDON 180 1.30 103 1.4 121 5.1 125 1.2 50 7.2 78 (7.9) LONG BEACH 341 (96.28) 92 2.1 188 0.9 105 3.3 136 (10.7) 95 (31.9) LOS ANGELES 337 (81.67) — 203 (0.4) 173 (7.8) 120 (3.8) 92 (24.0) The Container Port Performance Index 2020 92 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points LUANDA 343 (114.84) 116 0.3 297 (13.0) 204 (21.8) 160 (34.7) — LYTTELTON 186 0.08 — 162 2.9 147 (1.8) — — MAGDALLA 45 42.39 — 77 7.4 69 7.6 14 11.8 — MALABO 254 (13.57) — 298 (13.6) — — — MALAGA 129 9.91 105 1.3 129 4.8 111 2.9 — — MANAUS 270 (17.99) — 306 (18.0) — — — MANILA 276 (19.33) 202 (9.1) 247 (4.2) 169 (7.1) — — MANZANILLO 145 7.84 214 (13.9) 79 7.3 130 0.8 86 3.3 63 (0.3) (MEX-ICO) MAPUTO 306 (33.45) — 280 (9.6) 193 (15.5) — — MARIEL 197 (0.88) 158 (1.9) — — — — MARSAXLOKK 101 16.47 166 (3.1) 236 (3.3) 108 3.1 68 4.8 55 2.7 MARSEILLE 350 (238.49) 141 (1.2) 205 (0.4) 180 (9.9) 153 (24.6) 99 (67.7) MATADI 205 (2.28) 184 (4.9) — — — — MAWAN 29 55.67 — 24 10.5 2 13.2 12 12.6 — MEJILLONES 311 (38.26) — 65 7.9 87 5.3 141 (14.0) 84 (10.4) MELBOURNE 313 (40.21) — 238 (3.4) 159 (5.0) 143 (14.7) — MERSIN 83 25.30 95 2.0 185 1.1 114 2.7 96 1.7 37 6.1 MIAMI 114 13.96 61 3.7 136 4.2 110 3.0 94 1.7 — MOBILE 132 9.73 — 142 4.0 90 4.8 111 (0.9) — MOGADISCIO 260 (15.09) 186 (5.2) 293 (12.7) — — — MOJI 123 10.71 — 21 10.7 — — — MOMBASA 335 (78.90) 187 (5.3) 278 (9.3) 210 (43.6) — — MONTEVIDEO 328 (65.15) 39 5.1 155 3.5 185 (12.3) 155 (26.4) — MONTREAL 283 (21.59) — 315 (21.6) — — — MUHAMMAD 316 (42.02) — 237 (3.4) 168 (6.6) 142 (14.4) — BIN QASIM MUNDRA 44 42.74 — 20 10.7 62 8.2 27 9.9 — A Comparable Assessment of Container Port Performance 93 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points MUUGA— 198 (0.99) — 212 (1.0) — — — PORT OF TALLINN NAGOYA 48 41.34 5 8.7 7 11.8 58 8.4 57 6.4 — NAHA 111 14.65 — — 33 9.5 — — NANTES– 246 (10.26) 171 (3.9) 271 (8.5) — — — ST NAZAIRE NAPIER 261 (15.25) — 191 0.7 182 (10.3) — — NAPLES 207 (2.49) 20 6.7 40 9.0 179 (9.5) — — NASSAU 189 (0.14) 124 (0.3) — — — — NELSON 206 (2.46) — 228 (2.5) — — — NEMRUT BAY 326 (60.83) 74 3.2 141 4.0 122 1.5 — 94 (26.7) NEW 215 (4.47) 208 (9.7) — — — — MANGALORE NEW ORLEANS 237 (8.60) — 182 1.3 158 (5.0) 113 (1.1) — NEW YORK & 127 10.44 54 4.0 115 5.5 97 4.3 108 0.2 70 (1.5) NEW JERSEY NGQURA 349 (190.20) 159 (2.0) 279 (9.5) 198 (17.1) 152 (24.3) 97 (41.1) NINGBO 13 83.03 27 5.9 33 9.7 19 10.5 22 10.7 12 13.0 NORRKOPING 135 9.14 41 4.7 88 7.0 — — — NOUAKCHOTT 292 (25.32) 215 (14.6) 310 (18.6) — — — NOUMEA 124 10.63 31 5.5 57 8.1 — — — NOVOROSSIYSK 301 (30.45) 162 (2.2) 263 (7.1) 190 (14.5) — — OAKLAND 334 (78.72) 156 (1.8) 175 1.8 167 (6.3) 137 (12.0) 89 (18.0) ODESSA 265 (16.61) 13 7.8 145 3.9 — 138 (12.2) — OITA 188 (0.08) 122 (0.2) — — — — OMAEZAKI 121 10.99 — 16 11.0 — — — ONNE 291 (24.94) — 317 (24.9) — — — OSAKA 77 26.92 — 34 9.6 11 11.3 — — The Container Port Performance Index 2020 94 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points OSLO 91 20.15 82 2.8 1 18.9 — — — OTAGO 288 (23.73) — 166 2.4 197 (17.0) — — HARBOUR OWENDO 271 (18.43) 212 (11.9) 296 (12.9) — — — PAITA 94 19.43 — 100 6.2 55 8.6 — — PALERMO 182 0.77 101 1.7 — — — — PANJANG 229 (6.92) 188 (5.4) 248 (4.4) — — — PAPEETE 167 3.64 78 3.0 169 2.3 — — — PARANAGUA 107 15.52 — 195 (0.0) 113 2.7 63 5.8 — PECEM 86 23.27 — 76 7.6 107 3.2 64 5.4 — PENANG 227 (6.82) — 184 1.1 161 (5.2) — — PHILADELPHIA 87 23.11 — 52 8.5 34 9.5 — — PHILIPSBURG 192 (0.27) 102 1.5 211 (1.0) — — — PIPAVAV 39 48.40 — 9 11.6 4 12.6 37 8.8 — PIRAEUS 89 22.98 175 (4.3) 168 2.3 98 4.2 92 2.2 44 4.6 POINT LISAS 177 1.62 66 3.5 — — — — PORTS POINTE-A-PITRE 113 14.42 128 (0.4) 96 6.5 88 5.3 — — POINTE-NOIRE 345 (128.49) 218 (25.1) 304 (16.5) 207 (29.8) 157 (27.7) — PORT AKDENIZ 154 6.32 29 5.8 151 3.7 — — — PORT BOTANY 327 (62.93) 185 (5.0) 235 (3.1) 189 (13.1) 147 (19.0) — PORT BRONKA 152 6.60 69 3.5 124 5.0 — — — PORT 348 (182.71) — 274 (8.8) 208 (33.0) 162 (62.5) — ELIZABETH PORT 147 7.63 98 1.8 154 3.5 118 2.1 — — EVERGLADES PORT FREEPORT 193 (0.35) —201 (0.3) — — — PORT KLANG 14 78.19 56 3.9 84 7.1 28 9.9 18 11.3 16 12.4 PORT LOUIS 346 (175.17) 211 (11.4) 291 (12.0) 200 (17.4) 159 (31.9) 93 (26.6) A Comparable Assessment of Container Port Performance 95 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points PORT MORESBY 211 (3.60) 110 0.5 243 (3.8) — — — PORT OF SPAIN 248 (10.58) 198 (7.6) 262 (7.1) — — — PORT OF 110 14.77 133 (0.6) 127 4.9 120 2.0 77 4.1 64 (0.3) VIRGINIA PORT REUNION 340 (89.43) 192 (6.4) 285 (10.7) 195 (16.2) 154 (25.8) — PORT SAID 70 29.08 130 (0.5) 104 6.2 82 5.7 117 (2.5) 32 7.5 PORT VICTORIA 299 (28.31) — 322 (28.3) — — — POSORJA 126 10.50 — 36 9.5 — 103 0.5 — POTI 209 (2.69) 190 (5.8) — — — — PRINCE RUPERT 323 (56.06) — — 202 (18.9) 140 (13.7) — PUERTO 122 10.77 42 4.7 50 8.6 — — — BARRIOS PUERTO 155 6.15 — 105 6.2 — — — BOLIVAR (ECUADOR) PUERTO 307 (33.87) — 325 (33.9) — — — CABELLO PUERTO 149 7.16 93 2.1 102 6.2 — — — CORTES PUERTO LIMON 106 15.63 — 78 7.4 86 5.4 — — PUERTO 236 (8.43) 195 (7.0) 253 (5.2) — — — PROGRESO PUERTO 174 2.38 — 158 3.1 140 (0.8) 107 0.2 — QUETZAL QINGDAO 3 102.37 — 29 9.9 1 14.3 3 15.3 7 15.7 QINZHOU 148 7.57 150 (1.7) 54 8.3 — — — QUY NHON 104 16.14 33 5.4 3 13.7 — — — RAUMA 208 (2.65) 123 (0.2) 229 (2.6) — — — RAVENNA 176 1.93 40 4.8 197 (0.3) — — — RIGA 223 (5.97) 153 (1.7) 252 (5.2) — — — The Container Port Performance Index 2020 96 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points RIJEKA 279 (19.51) 77 3.0 138 4.1 150 (1.9) — 81 (8.6) RIO DE JANEIRO 309 (34.73) — 69 7.8 116 2.3 151 (23.4) — RIO GRANDE 47 41.54 — 46 8.8 36 9.4 35 9.3 — (BRAZIL) RIO HAINA 160 5.14 94 2.0 137 4.2 — — — ROTTERDAM 90 22.41 180 (4.7) 176 1.8 91 4.8 84 3.4 50 3.4 SAIGON 151 6.94 — 89 6.9 — — — SAINT JOHN 251 (11.59) 120 0.1 286 (10.9) 136 (0.5) — — SALALAH 9 87.45 144 (1.3) 13 11.4 24 10.3 6 13.1 11 13.6 SALERNO 168 3.44 137 (0.9) 147 3.9 — — — SALVADOR 55 35.29 — 47 8.7 93 4.6 28 9.9 — SAMSUN 280 (19.87) 205 (9.5) 303 (15.5) — — — SAN ANTONIO 173 2.53 — 224 (2.1) 155 (4.4) 116 (2.5) 36 6.3 SAN JUAN 164 4.60 100 1.7 148 3.8 — — — SAN PEDRO 293 (27.01) — 318 (27.0) — — — (COTE D'IVOIRE) SAN VICENTE 226 (6.50) — 173 1.9 162 (5.4) — — SANTA CRUZ DE 95 18.66 65 3.6 196 (0.2) — 41 8.7 — TENERIFE SANTA MARTA 137 8.66 — 49 8.7 — — — SANTO TOMAS 201 (1.71) — 221 (1.7) — — — DE CASTILLA SANTOS 72 28.25 — 123 5.1 60 8.3 66 5.3 — SAVANNAH 96 18.19 108 1.1 106 6.1 94 4.5 79 3.6 68 (1.0) SAVONA-VADO 314 (40.38) 155 (1.8) 204 (0.4) 175 (8.1) 139 (13.6) — SEATTLE 263 (16.29) — 206 (0.4) 157 (4.8) 123 (4.3) — SEPETIBA 84 24.32 — 25 10.5 74 6.8 95 1.7 — SHANGHAI 49 41.30 7 8.5 19 10.7 17 10.7 25 10.3 71 (3.9) A Comparable Assessment of Container Port Performance 97 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points SHANTOU 117 12.02 — 87 7.0 106 3.3 — — SHEKOU 5 93.69 21 6.6 43 8.9 14 10.9 16 11.7 5 16.3 SHIBUSHI 166 3.71 12 8.1 — — — — SHIMIZU 24 59.39 4 9.5 17 10.9 7 12.2 7 12.9 — SHUAIBA 228 (6.89) — 260 (6.9) — — — SHUWAIKH 199 (1.32) — 216 (1.3) — — — SIAM SEAPORT 136 8.72 — 48 8.7 — — — SINES 28 56.19 — 37 9.3 99 4.0 30 9.7 29 8.4 SINGAPORE 12 83.12 52 4.1 53 8.4 32 9.5 38 8.8 6 15.9 SOHAR 69 29.49 — 159 3.0 37 9.2 54 7.0 66 (0.6) SOKHNA 321 (54.65) — 165 2.4 196 (16.8) 121 (4.0) 82 (9.1) SOUTHAMPTON 317 (44.52) 142 (1.2) 266 (7.6) 124 1.2 128 (6.1) 83 (10.2) ST PETERSBURG 258 (14.24) 181 (4.7) 292 (12.1) — — — STOCKHOLM 253 (13.50) 203 (9.3) 276 (9.2) — — — SUAPE 130 9.88 — 101 6.2 121 1.8 105 0.4 — TACOMA 329 (66.41) — 283 (10.4) 192 (14.6) 145 (17.1) — TAICHUNG 146 7.74 18 7.0 133 4.5 — — — TAIPEI, TAIWAN 16 75.41 — — — 2 16.6 4 16.6 CHINA TAKORADI 235 (7.89) 132 (0.6) 268 (7.6) — — — TAMPA 200 (1.40) — 110 5.7 128 1.1 124 (4.4) — TANGER MEDI- 15 76.09 45 4.4 82 7.2 21 10.4 31 9.6 15 12.4 TERRANEAN TANJUNG EMAS 143 7.91 — 64 7.9 — — — TANJUNG 11 85.62 64 3.7 74 7.7 26 10.1 23 10.5 8 15.5 PELEPAS TANJUNG 172 2.88 — 144 3.9 139 (0.7) — — PERAK TANJUNG PRIOK 66 30.57 138 (1.0) 126 4.9 76 6.3 87 3.3 48 3.9 The Container Port Performance Index 2020 98 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points TARRAGONA 134 9.19 47 4.4 83 7.2 — — — TAURANGA 100 16.52 216 (17.6) 81 7.3 96 4.4 65 5.4 — TEESPORT 238 (8.61) 165 (2.5) 265 (7.4) — — — TEMA 312 (39.91) — 256 (5.9) 174 (7.9) 109 (0.3) 80 (8.3) THESSALONIKI 284 (21.76) 179 (4.6) 251 (5.1) 178 (9.5) — — TIANJIN 20 64.01 199 (7.8) 97 6.5 42 9.0 33 9.5 19 11.1 TIMARU 304 (31.50) — 219 (1.5) 203 (19.5) — — TIN CAN ISLAND 344 (117.73) — 323 (29.2) 213 (57.5) — — TOAMASINA 219 (5.04) — 250 (5.0) — — — TOKYO 54 37.68 50 4.2 58 8.1 46 8.9 52 7.1 — TOMAKOMAI 204 (1.93) 173 (4.2) — — — — TRIESTE 264 (16.41) 90 2.3 177 1.7 126 1.1 — 79 (8.1) TRIPOLI 76 26.93 6 8.5 181 1.3 53 8.6 75 4.3 — (LEBANON) TUTICORIN 139 8.29 30 5.6 111 5.7 — — — ULSAN 175 1.97 — 125 5.0 151 (1.9) — — UMM QASR 294 (27.28) — 290 (12.0) 181 (9.9) — — VALENCIA 308 (34.16) 201 (8.8) 239 (3.6) 170 (7.1) 104 0.5 77 (6.4) VALPARAISO 102 16.25 — 128 4.8 44 8.9 114 (1.2) — VANCOUVER 303 (30.98) — 245 (3.9) 172 (7.6) 130 (7.8) — (CAN-ADA) VARNA 213 (3.78) 161 (2.1) 232 (2.8) — — — VENICE 295 (27.37) 160 (2.0) 246 (4.0) 191 (14.6) — — VERACRUZ 97 18.16 113 0.3 116 5.4 63 8.2 — — VIGO 159 5.36 37 5.2 160 3.0 — — — VILA DO CONDE 241 (9.24) — 277 (9.2) — — — VITORIA 218 (5.03) — 249 (5.0) — — — VLADIVOSTOK 240 (9.11) 172 (4.0) 264 (7.3) — — — A Comparable Assessment of Container Port Performance 99 Weight 0.46 1.00 1.54 1.97 2.57 <1,500 1,501–5,000 5,001–8,500 8,501–13,500 >13,500 TEU TEU TEU TEU TEU Port name Rank Total Rank Points Rank Points Rank Points Rank Points Rank Points points VOSTOCHNY 255 (13.76) 189 (5.8) 288 (11.1) — — — WALVIS BAY 336 (79.59) 182 (4.7) 295 (12.9) 199 (17.2) 148 (19.3) — WELLINGTON 141 8.12 — 107 6.0 123 1.4 — — WILHELM- 52 38.84 35 5.3 86 7.0 71 7.3 58 6.2 56 2.3 SHAVEN WILMINGTON 67 30.06 — 59 8.1 31 9.7 81 3.6 — (NORTH CAROLINA, USA) XIAMEN 26 58.04 49 4.2 32 9.7 41 9.0 29 9.7 41 5.2 YANGSHAN 10 86.53 3 9.7 8 11.6 15 10.8 24 10.5 13 12.9 YANTIAN 17 72.53 25 6.2 61 8.1 43 9.0 39 8.8 17 11.9 YARIMCA 58 33.01 68 3.5 234 (3.1) 92 4.7 53 7.0 40 5.2 YOKKAICHI 81 25.63 — 4 12.2 51 8.8 — — YOKOHAMA 1 129.74 23 6.5 26 10.5 3 12.8 1 18.4 1 23.5 YEOSU 40 48.08 67 3.5 41 9.0 20 10.4 42 8.5 59 1.8 YUZHNY 170 3.25 — 122 5.1 143 (1.2) — — ZEEBRUGGE 225 (6.41) 16 7.3 186 1.0 — 102 0.9 74 (4.9) ZHOUSHAN 8 87.63 — 23 10.7 23 10.3 11 12.6 9 14.1 Source: Original calculations for this publication, based on CPPI 2020 data. Note: TEU = twenty-foot equivalent unit The Container Port Performance Index 2020 100 Table A.2. The CPPI 2020: Statistical Approach Port name Rank Total score Port name Rank Total score AARHUS 44 −2.036 BEIRUT 11 −3.378 ABIDJAN 228 0.439 BEJAIA 338 4.054 ACAJUTLA 249 0.640 BELAWAN 197 0.104 ADELAIDE 339 4.546 BELL BAY 163 −0.205 AGADIR 276 1.122 BERBERA 223 0.368 AGUADULCE (COLOMBIA) 45 −2.035 BILBAO 224 0.384 ALEXANDRIA (EGYPT) 234 0.482 BLUFF 257 0.731 ALGECIRAS 10 −3.597 BORUSAN 166 −0.163 ALGIERS 315 2.076 BOSTON (USA) 113 −0.631 ALTAMIRA 145 −0.324 BREMERHAVEN 41 −2.265 AMBARLI 57 −1.783 BRISBANE 246 0.569 ANCONA 193 0.067 BRISTOL 294 1.462 ANTWERP 86 −1.011 BUENAVENTURA 72 −1.353 APRA HARBOR 248 0.610 BUENOS AIRES 320 2.391 AQABA 60 −1.594 BURGAS 117 −0.611 ARICA 207 0.211 BUSAN 52 −1.887 ASHDOD 327 2.797 CAGAYAN DE ORO 122 −0.582 AUCKLAND 123 −0.562 CAI LAN 46 −1.991 BALBOA 114 −0.625 CAI MEP 49 −1.932 BALTIMORE (USA) 138 −0.368 CALDERA (COSTA RICA) 162 −0.206 BANGKOK 271 1.024 CALLAO 231 0.450 BARCELONA 76 −1.224 CAPE TOWN 347 6.528 BARI 201 0.167 CARTAGENA (COLOMBIA) 42 −2.185 BARRANQUILLA 169 −0.137 CARTAGENA (SPAIN) 168 −0.153 BATANGAS 112 −0.663 CASABLANCA 291 1.442 BATUMI 316 2.175 CASTELLON 144 −0.339 BEIRA 268 0.919 CAT LAI 80 −1.149 A Comparable Assessment of Container Port Performance 101 Port name Rank Total score Port name Rank Total score CATANIA 180 −0.021 DURBAN 349 8.082 CAUCEDO 40 −2.355 DURRES 190 0.057 CEBU 151 −0.275 DUTCH HARBOR 325 2.591 CHARLESTON 95 −0.820 EL DEKHEILA 239 0.509 CHATTOGRAM 306 1.809 ENSENADA 136 −0.378 CHIBA 134 −0.418 FELIXSTOWE 313 2.006 CHIWAN 3 −5.202 FORT-DE-FRANCE 125 −0.543 CHORNOMORSK 230 0.449 FREDERICIA 128 −0.493 CIVITAVECCHIA 175 −0.072 FREEPORT (BAHAMAS) 152 −0.271 COLOMBO 17 −3.209 FREETOWN 220 0.350 COLON 100 −0.752 FREMANTLE 326 2.716 CONAKRY 238 0.505 FUZHOU 23 −2.829 CONSTANTZA 253 0.660 GAVLE 213 0.242 COPENHAGEN 129 −0.471 GDANSK 311 1.944 CORONEL 282 1.203 GDYNIA 90 −0.942 COTONOU 244 0.562 GEMLIK 51 −1.892 CRISTOBAL 170 −0.135 GENERAL SANTOS 278 1.148 DA CHAN BAY TERMINAL ONE 31 −2.588 GENOA 321 2.420 DAKAR 177 −0.031 GIJON 187 0.013 DALIAN 33 −2.506 GIOIA TAURO 142 −0.344 DAMIETTA 250 0.650 GOTHENBURG 341 4.653 DAMMAM 102 −0.737 GRANGEMOUTH 243 0.560 DANANG 84 −1.053 GUANGZHOU 4 −5.162 DAR ES SALAAM 324 2.561 GUAYAQUIL 336 3.647 DAVAO 143 −0.341 GUSTAVIA 126 −0.528 DILISKELESI 54 −1.842 HAIFA 225 0.384 DJIBOUTI 61 −1.590 HAIPHONG 47 −1.953 DOUALA 298 1.510 HAKATA 99 −0.758 DUBLIN 262 0.839 HALIFAX 39 −2.365 DUNKIRK 118 −0.611 HAMAD PORT 36 −2.411 The Container Port Performance Index 2020 102 Port name Rank Total score Port name Rank Total score HAMBURG 280 1.176 KRIBI DEEP SEA PORT 303 1.701 HELSINGBORG 164 −0.196 KRISTIANSAND 186 0.008 HERAKLION 178 −0.030 LA GUAIRA 305 1.737 HONG KONG PORT, 7 −4.276 LA SPEZIA 241 0.548 HONG KONG SAR, CHINA LAE 287 1.314 HOUSTON 266 0.878 LAEM CHABANG 55 −1.807 HUENEME 264 0.862 LAGOS (NIGERIA) 340 4.646 INCHEON 34 −2.422 LARVIK 181 −0.015 IQUIQUE 342 4.766 LATAKIA 171 −0.135 ISKENDERUN 272 1.024 LAZARO CARDENAS 25 −2.798 ITAJAI 308 1.828 LE HAVRE 258 0.746 ITAPOA 69 −1.376 LEIXOES 206 0.201 IZMIR 277 1.136 LIANYUNGANG 37 −2.375 JACKSONVILLE 161 −0.215 LIMASSOL 192 0.063 JAWAHARLAL NEHRU PORT 56 −1.786 LIRQUEN 256 0.708 JEBEL ALI 323 2.482 LIVORNO 251 0.658 JEDDAH 53 −1.862 LOME 288 1.332 JOHOR 110 −0.669 LONDON 82 −1.117 JUBAIL 21 −2.898 LONG BEACH 333 3.175 KALININGRAD 167 −0.159 LOS ANGELES 328 2.899 KAOHSIUNG 5 −4.669 LUANDA 350 8.383 KARACHI 75 −1.292 LYTTELTON 188 0.034 KEELUNG 65 −1.509 MAGDALLA 48 −1.943 KHALIFA BIN SALMAN 140 −0.356 MALABO 263 0.859 KHALIFA PORT 26 −2.795 MALAGA 108 −0.690 KING ABDULLAH PORT 2 −5.684 MANAUS 269 0.967 KINGSTON (JAMAICA) 286 1.283 MANILA 322 2.445 KLAIPEDA 301 1.635 MANZANILLO (MEXICO) 275 1.070 KOBE 19 −3.127 MAPUTO 299 1.533 KOPER 318 2.237 MARIEL 191 0.057 KOTKA 242 0.558 MARSAXLOKK 106 −0.698 A Comparable Assessment of Container Port Performance 103 Port name Rank Total score Port name Rank Total score MARSEILLE 345 5.696 NOUAKCHOTT 295 1.475 MATADI 209 0.238 NOUMEA 104 −0.721 MAWAN 32 −2.557 NOVOROSSIYSK 300 1.626 MEJILLONES 289 1.381 OAKLAND 332 3.163 MELBOURNE 302 1.676 ODESSA 189 0.046 MERSIN 214 0.275 OITA 185 0.004 MIAMI 200 0.149 OMAEZAKI 109 −0.680 MOBILE 101 −0.745 ONNE 270 0.994 MOGADISCIO 281 1.194 OSAKA 67 −1.440 MOJI 111 −0.663 OSLO 78 −1.192 MOMBASA 331 3.140 OTAGO HARBOUR 314 2.023 MONTEVIDEO 273 1.033 OWENDO 245 0.563 MONTREAL 285 1.231 PAITA 105 −0.705 MUHAMMAD BIN QASIM 296 1.499 PALERMO 172 −0.125 MUNDRA 50 −1.902 PANJANG 247 0.573 MUUGA—PORT OF TALLINN 195 0.092 PAPEETE 157 −0.227 NAGOYA 16 −3.251 PARANAGUA 252 0.659 NAHA 92 −0.883 PECEM 59 −1.647 NANTES−ST NAZAIRE 255 0.693 PENANG 139 −0.367 NAPIER 240 0.513 PHILADELPHIA 83 −1.080 NAPLES 150 −0.280 PHILIPSBURG 194 0.090 NASSAU 173 −0.109 PIPAVAV 38 −2.371 NELSON 210 0.240 PIRAEUS 93 −0.859 NEMRUT BAY 329 2.970 POINT LISAS PORTS 165 −0.182 NEW MANGALORE 232 0.451 POINTE-A-PITRE 147 −0.319 NEW ORLEANS 265 0.865 POINTE-NOIRE 346 5.832 NEW YORK & NEW JERSEY 89 −0.969 PORT AKDENIZ 96 −0.820 NGQURA 351 8.401 PORT BOTANY 337 3.907 NINGBO 24 −2.805 PORT BRONKA 132 −0.454 NORRKOPING 124 −0.553 PORT ELIZABETH 348 7.659 The Container Port Performance Index 2020 104 Port name Rank Total score Port name Rank Total score PORT EVERGLADES 208 0.213 ROTTERDAM 116 −0.611 PORT FREEPORT 182 −0.008 SAIGON 137 −0.375 PORT KLANG 14 −3.334 SAINT JOHN 211 0.241 PORT LOUIS 344 5.501 SALALAH 6 −4.531 PORT MORESBY 217 0.307 SALERNO 174 −0.080 PORT OF SPAIN 260 0.775 SALVADOR 43 −2.051 PORT OF VIRGINIA 85 −1.044 SAMSUN 297 1.502 PORT REUNION 334 3.302 SAN ANTONIO 159 −0.218 PORT SAID 58 −1.652 SAN JUAN 148 −0.303 PORT VICTORIA 292 1.457 SAN PEDRO (COTE D'IVOIRE) 319 2.267 POSORJA 77 −1.192 SAN VICENTE 221 0.361 POTI 215 0.278 SANTA CRUZ DE TENERIFE 88 −0.970 PRINCE RUPERT 330 2.979 SANTA MARTA 120 −0.589 PUERTO BARRIOS 103 −0.731 SANTO TOMAS DE CASTILLA 203 0.185 PUERTO BOLIVAR (ECUADOR) 141 −0.347 SANTOS 70 −1.376 PUERTO CABELLO 309 1.829 SAVANNAH 279 1.158 PUERTO CORTES 131 −0.468 SAVONA-VADO 310 1.897 PUERTO LIMON 94 −0.858 SEATTLE 236 0.487 PUERTO PROGRESO 259 0.750 SEPETIBA 73 −1.338 PUERTO QUETZAL 154 −0.269 SHANGHAI 63 −1.532 QINGDAO 8 −3.860 SHANTOU 91 −0.935 QINZHOU 127 −0.523 SHEKOU 9 −3.726 QUY NHON 79 −1.163 SHIBUSHI 135 −0.382 RAUMA 229 0.439 SHIMIZU 12 −3.361 RAVENNA 179 −0.025 SHUAIBA 149 −0.301 RIGA 235 0.482 SHUWAIKH 160 −0.216 RIJEKA 312 1.974 SIAM SEAPORT 115 −0.613 RIO DE JANEIRO 199 0.126 SINES 18 −3.183 RIO GRANDE (BRAZIL) 74 −1.332 SINGAPORE 15 −3.279 RIO HAINA 146 −0.322 SOHAR 71 −1.375 A Comparable Assessment of Container Port Performance 105 Port name Rank Total score Port name Rank Total score SOKHNA 293 1.459 TUTICORIN 130 −0.469 SOUTHAMPTON 290 1.404 ULSAN 98 −0.811 ST PETERSBURG 304 1.719 UMM QASR 233 0.452 STOCKHOLM 267 0.915 VALENCIA 283 1.211 SUAPE 81 −1.129 VALPARAISO 222 0.365 TACOMA 335 3.628 VANCOUVER (CANADA) 237 0.499 TAICHUNG 97 −0.814 VARNA 196 0.094 TAIPEI, TAIWAN, CHINA 30 −2.681 VENICE 274 1.042 TAKORADI 254 0.683 VERACRUZ 158 −0.219 TAMPA 226 0.411 VIGO 156 −0.233 TANGER MEDITERRANEAN 27 −2.769 VILA DO CONDE 183 0.000 TANJUNG EMAS 133 −0.433 VITORIA 183 0.000 TANJUNG PELEPAS 13 −3.342 VLADIVOSTOK 202 0.182 TANJUNG PERAK 153 −0.269 VOSTOCHNY 204 0.192 TANJUNG PRIOK 64 −1.521 WALVIS BAY 307 1.819 TARRAGONA 121 −0.588 WELLINGTON 176 −0.049 TAURANGA 219 0.333 WILHELMSHAVEN 119 −0.598 TEESPORT 216 0.303 WILMINGTON (NORTH CAROLINA, USA) 155 −0.237 TEMA 261 0.782 XIAMEN 62 −1.541 THESSALONIKI 284 1.229 YANGSHAN 28 −2.733 TIANJIN 218 0.310 YANTIAN 29 −2.724 TIMARU 317 2.225 YARIMCA 68 −1.393 TIN CAN ISLAND 343 4.789 YOKKAICHI 107 −0.692 TOAMASINA 227 0.433 YOKOHAMA 1 −5.995 TOKYO 35 −2.418 YEOSU 22 −2.831 TOMAKOMAI 205 0.194 YUZHNY 198 0.105 TRIESTE 212 0.241 ZEEBRUGGE 87 −0.988 TRIPOLI (LEBANON) 66 −1.497 ZHOUSHAN 20 −2.963 Source: Original calculations for this publication, based on CPPI 2020 data. The Container Port Performance Index 2020 106 Table A.3. The Ports Omitted from CPPI 2020 Port name Port name Port name TARTOUS CHENNAI BRIDGETOWN TEKIRDAG LAS PALMAS PORTSMOUTH (UNITED KINGDOM) HOSOSHIMA GEORGETOWN (GUYANA) YANGON KAWASAKI MAMONAL ST GEORGE'S (GRENADA) MELILLA CADIZ PARAMARIBO COCHIN APIA ZHANJIANG SAN CARLOS (PHILIPPINES) BAR QASR AHMED VISAKHAPATNAM PRACHUAP PORT CASTRIES BARCADERA CORK BORDEAUX MANTA PORT-AU-PRINCE RADES POZZALLO MAZATLAN DIEGO SUAREZ KAMARAJAR PLOCE WILLEMSTAD HELSINKI LA ROCHELLE-PALLICE SKIKDA KATTUPALLI HONOLULU GREENOCK KRISHNAPATNAM ALICANTE TURBO SHARJAH CHU LAI MONROVIA TRAPANI NGHI SON TRABZON KOMPONG SOM ANTOFAGASTA BENGHAZI PUERTO BOLIVAR (COLOMBIA) LISBON FORTALEZA SUBIC BAY LIVERPOOL (UNITED KINGDOM) ESMERALDAS KUANTAN KING FAHD INDUSTRIAL PORT (YANBU) VLISSINGEN UST–LUGA BREST PAGO-PAGO WILMINGTON (DELAWARE, USA) VENTERMINALES DR SYAMA PRASAD MOOKERJEE CHONGMING TOWNSVILLE MAYOTTE MARIN (SPAIN) CORINTO ADEN Source: IHS Markit, based on CPPI 2020 data. A Comparable Assessment of Container Port Performance 107 Appendix B: Constructing the CPPI with Factor Analysis Structure of the data 103. Before discussing the methodology employed in constructing the CPPI with factor analysis (FA), it is helpful to first summarize the structure of available data. The dataset contains following five cate- gories of ship size: a. Feeders: <1,500 TEU b. Intraregional: 1,500–5,000 TEU c. Intermediate: 5,000–8,500 TEU d. Neo-Panamax: 8,500–13,500 TEU e. Ultra-large container carriers: >13,500 TEU 104. For each category, there are ten different bands for call size. The port productivity is captured by average idle hour, which consists of two parts: port-to-berth (PB) and on-berth (B). Thus, ideally, the total variables = 5 × 10 × 2. Of course, many of them have missing values. The objective is to build a factor model to summarize these variables and then construct a port productivity index for all ports under con- sideration. The data structure for a particular category k (k = 1, 2, 3, 4, 5) can be summarized as shown in table B.1. In the following, we use j = 1, 2, . . . to denote various bands of call size and i = 1, 2, . . . to denote different ports under consideration. The Container Port Performance Index 2020 108 Table B.1. Sample Port Productivity Data Structure, by Ship Size Ship size (k) Call size band (number of moves) <250 251–500 ... >6,000 Ports Port-to- Berth Port-to- Berth Port-to- Berth berth berth berth 1 2 3 ... Imputation of missing values 105. A major practical problem is that most idle hour variables have a significant number of missing values. For instance, the port perfor- mance dataset includes very limited observations for large call size bands when ship capacity is low. Those columns with no data can be just deleted, as they convey no information. One way to deal with missing data in variables is to use only complete cases when extract- ing factors. This, however, will dramatically reduce the sample size and cause information loss. 106. A more sophisticated approach is to use likelihood-based method to impute those missing values. For the current dataset, expecta- tion–maximization (EM) algorithm can be utilized to provide a max- imum-likelihood estimator for each missing value. It relies on two critical assumptions: The first assumption is that gaps are random, or more specifically, the gaps are not caused by sample selection bias. The second assumption is that all variables under consideration fol- low a normal distribution. Given the dataset, these two assumptions are plausible. EM computes the maximum likelihood estimator for the mean and variance of the normal distribution given the observed data. Knowing the distribution that generates the missing data, we can then sample from it to impute the missing values. FA can then be based on the resultant complete dataset, instead of the original one filled with missing values. A subtle point to note: The EM algorithm could produce a negative value for some missing data, which makes little sense as idle hours cannot be less than zero. If a negative value occurs, the value should be set to zero. A Comparable Assessment of Container Port Performance 109 107. The proposed imputation method relies on some randomness, spe- cifically, the missing values are filled by drawing randomly from a normal distribution. This causes a lot of uncertainty, as an extreme value can be drawn by chance. Despite a very low probability, it could happen if drawn only once from the dataset. Thus, the results based on this kind of imputation could be quite unstable. The following can be done to overcome this problem: 1. Impute the missing data, say, 1,000 times (each time will produce a different “complete dataset” due to the randomness); 2. Take the average over those 1,000 complete datasets to obtain final imputation; and 3. Run the FA using this dataset and obtain the final rankings. 108. Using this approach, we find the scores produced are much more stable and closer to the prior expectation. After thorough testing, 1,000 replications proved sufficient to provide a satisfactory result in all cases. It could be posited that, as the algorithm involves random- ness, different results could be produced every time. (The missing value is dealt with by the “norm” package in the statistical software used (R,) which employs an EM imputation algorithm). That would not be a problem, however, as the so-called “random seed” can be set exogenously, which allows “control” of the random sequence gener- ated by the computer. Intuitively, if the same seed—which is noth- ing more than a positive integer—is used, then the same conclusion should be reached. Why is factor analysis useful? 109. Essentially, for each port, quite a few variables contain information about its efficiency. These include average time cost under various categories: (i) different call size bands, and (ii) berth/port-to-berth. The reason FA can be helpful is these variables are in fact determined by a small number of unobserved factors, which might include quality of infrastructure, expertise of staff, and so on. Usually, it means just two or three such factors can summarize almost all useful informa- tion. The challenge lies in the inability to observe those latent factors; however, a simple example could be helpful: Imagine three ports, each with four different types of time cost, as shown in table B.2. The Container Port Performance Index 2020 110 Table B.2. Simple Illustration of Latent Factors Port Cost 1 Cost 2 Cost 3 Cost 4 A 1 2 3 4 B 2 4 6 8 C 3 6 9 12 110. As observed in table B.2, costs 2, 3, and 4 are simply multiples of cost 1. Although each port has four variables, one variable alone is enough to rank the efficiency of these three ports (A > B > C). Of course, this example represents an extreme case, the idea can be generalized if these variables are correlated to a less extreme extent. In that case, the factors are computed as a linear combination of costs 1 through 4. However, if costs 1 through 4 are completely inde- pendent of each other, this method makes no sense. Fortunately, this is not the case for the CPPI dataset. Thus, for each port, we can com- pute its score on all factors and then combine those scores together to reach a final efficiency score. 111. Note that for FA, we do not calculate scores for each call size range. On the contrary, the whole dataset is used simultaneously to obtain latent factors. This contrasts sharply with the administrative method- ology. The approach using FA utilizes all correlations among hours for various call size bands, which from a statistical viewpoint is more efficient. One downside is that FA does not consider that some obser- vations are more reliable than others as they are based on more calls. Therefore, the result could be seriously distorted in the presence of large outliers. This is an area necessitating further investigation in future iterations of the CPPI. A Comparable Assessment of Container Port Performance 111 Statistical methodology 112. The final index is a weighted average of five subindices. Specifically, for a particular port i, it is defined as: $ !""#(%) = ( )! ∙ #!,# $ !%& !""#(%) = ( )! ∙ #!,# Where Ik,i denote the index for five 113. !%&ship size categories under con- sideration. The weights w_k should be chosen according to the relative importance of these five cases and intuitively it should sat- isfy w1